bigcomputer/kaggle-notebooks-outputs-filtered-44

text stringlengths 1.42k 99.1M	id stringlengths 21 24	file_path stringclasses 46 values
{ "cells": [ { "cell_type": "markdown", "id": "145de7c9", "metadata": { "papermill": { "duration": 0.038294, "end_time": "2021-08-24T18:33:41.181818", "exception": false, "start_time": "2021-08-24T18:33:41.143524", "status": "completed" }, "tags": [] }, "source": [ "# Problem Statement\n", "\n", "The head of HR of a certain organization wants to automate their salary hike estimation. The organization consulted an analytics service provider and asked them to build a basic prediction model by providing them with a dataset that contains the data about the number of years of experience and the salary hike given accordingly. Build a Simple Linear Regression model with salary as the target variable.\n", "\n", "Definition of Variable -:\n", "\n", "1. Year of Experience (Input)\n", "2. Salary Hike -> Hike in salary (Target)\n" ] }, { "cell_type": "markdown", "id": "e5afb14c", "metadata": { "papermill": { "duration": 0.036479, "end_time": "2021-08-24T18:33:41.256896", "exception": false, "start_time": "2021-08-24T18:33:41.220417", "status": "completed" }, "tags": [] }, "source": [ "# Approach to Solve the Business Case -:\n", "1. Business Objective - Find Relationship between Salary Hike and Year of Experience\n", "2. Perform EDA on data (Outlier, Missing Values).\n", "3. Understand the relationship between the variable's using Scatter Plot.\n", "4. Apply Simple linear regression with OLE to create base regression model( Vanilla Model)\n", "5. Check the RMSE, R^2 & R values for the model.\n", "5. Compare the Vanilla model with Model with Transformation's to check the best fit model.\n", " Transformation used\n", " 1. Logarithmic\n", " 2. Exponential\n", " 3. Polynomial with degree 2\n", "6. After comparing RMSE, R^2 & R values, Select the best model.\n", "7. Train & Test your data on these model to check the performance of model on test data.\n", "\n" ] }, { "cell_type": "code", "execution_count": 1, "id": "7c8a58fe", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:33:41.355760Z", "iopub.status.busy": "2021-08-24T18:33:41.345678Z", "iopub.status.idle": "2021-08-24T18:33:42.274231Z", "shell.execute_reply": "2021-08-24T18:33:42.272656Z", "shell.execute_reply.started": "2021-08-24T18:28:29.916029Z" }, "papermill": { "duration": 0.971429, "end_time": "2021-08-24T18:33:42.274538", "exception": true, "start_time": "2021-08-24T18:33:41.303109", "status": "failed" }, "tags": [] }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "UsageError: Line magic function `%` not found.\n" ] } ], "source": [ "#Import Library\n", "import pandas as pd\n", "import numpy as np\n", "import seaborn as sns\n", "import matplotlib.pyplot as plt\n", "% matplotlib inline\n", "import statsmodels.formula.api as smf\n" ] }, { "cell_type": "code", "execution_count": null, "id": "8181f09e", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:31.375993Z", "iopub.status.busy": "2021-08-24T18:28:31.375240Z", "iopub.status.idle": "2021-08-24T18:28:31.387388Z", "shell.execute_reply": "2021-08-24T18:28:31.386508Z", "shell.execute_reply.started": "2021-08-24T18:28:31.375936Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Import the dataset\n", "sal = pd.read_csv(\"../input/salary/Salary.csv\")" ] }, { "cell_type": "code", "execution_count": null, "id": "42db0d3d", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:33.377308Z", "iopub.status.busy": "2021-08-24T18:28:33.375351Z", "iopub.status.idle": "2021-08-24T18:28:33.387130Z", "shell.execute_reply": "2021-08-24T18:28:33.386152Z", "shell.execute_reply.started": "2021-08-24T18:28:33.377271Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Check the data\n", "sal.head()" ] }, { "cell_type": "markdown", "id": "fcd7a320", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "Change the Column Label to short Abbrevation for ease of Code handling.\n", "Weight gained (grams) - wg\n", "Calories Consumed - cc" ] }, { "cell_type": "markdown", "id": "b63802b6", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "### Rename Column for Ease of Coding" ] }, { "cell_type": "code", "execution_count": null, "id": "e326b80b", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:35.875585Z", "iopub.status.busy": "2021-08-24T18:28:35.875250Z", "iopub.status.idle": "2021-08-24T18:28:35.887145Z", "shell.execute_reply": "2021-08-24T18:28:35.886321Z", "shell.execute_reply.started": "2021-08-24T18:28:35.875553Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "sal = sal.rename(columns={\"Salary\":\"hike\",\"YearsExperience\":\"years\"})\n", "sal.head(5)\n" ] }, { "cell_type": "markdown", "id": "73141ddd", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "# 2. Perform EDA on data (Outlier, Missing Values)" ] }, { "cell_type": "markdown", "id": "9edccd42", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "#Exploratory Data Analysis\n", "1. Check for the null Values in dataset.\n", "2. Check for the outlier in dataset\n" ] }, { "cell_type": "code", "execution_count": null, "id": "b3c7ded2", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:39.015306Z", "iopub.status.busy": "2021-08-24T18:28:39.014992Z", "iopub.status.idle": "2021-08-24T18:28:39.029879Z", "shell.execute_reply": "2021-08-24T18:28:39.028852Z", "shell.execute_reply.started": "2021-08-24T18:28:39.015280Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Check for Null Data - No Null values\n", "sal.info()" ] }, { "cell_type": "code", "execution_count": null, "id": "c4369be2", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:41.355559Z", "iopub.status.busy": "2021-08-24T18:28:41.355231Z", "iopub.status.idle": "2021-08-24T18:28:41.373607Z", "shell.execute_reply": "2021-08-24T18:28:41.372617Z", "shell.execute_reply.started": "2021-08-24T18:28:41.355531Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "sal.describe()" ] }, { "cell_type": "code", "execution_count": null, "id": "c99b170a", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:43.395591Z", "iopub.status.busy": "2021-08-24T18:28:43.395282Z", "iopub.status.idle": "2021-08-24T18:28:43.594317Z", "shell.execute_reply": "2021-08-24T18:28:43.593701Z", "shell.execute_reply.started": "2021-08-24T18:28:43.395563Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Plot Histogram to view Distribution of data field ( Univariate)\n", "sns.histplot(sal[\"hike\"], color ='green',kde=True)" ] }, { "cell_type": "code", "execution_count": null, "id": "b8d583c5", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:45.890783Z", "iopub.status.busy": "2021-08-24T18:28:45.890295Z", "iopub.status.idle": "2021-08-24T18:28:46.063523Z", "shell.execute_reply": "2021-08-24T18:28:46.062873Z", "shell.execute_reply.started": "2021-08-24T18:28:45.890751Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Plot Histogram to view Distribution of data field ( Univariate) \n", "sns.histplot(sal[\"years\"], color ='blue',kde=True)" ] }, { "cell_type": "markdown", "id": "7bd0ca60", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "#Outlier Identification\n", "Use BoxPlot to Identify any outlier in data" ] }, { "cell_type": "code", "execution_count": null, "id": "79e7c86b", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:47.955671Z", "iopub.status.busy": "2021-08-24T18:28:47.955171Z", "iopub.status.idle": "2021-08-24T18:28:48.073727Z", "shell.execute_reply": "2021-08-24T18:28:48.072628Z", "shell.execute_reply.started": "2021-08-24T18:28:47.955641Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "sns.boxplot(sal[\"hike\"], color ='green')\n", "# No Outlier found based on the boxplot analysis" ] }, { "cell_type": "code", "execution_count": null, "id": "2a7a75a8", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:50.315537Z", "iopub.status.busy": "2021-08-24T18:28:50.315211Z", "iopub.status.idle": "2021-08-24T18:28:50.439779Z", "shell.execute_reply": "2021-08-24T18:28:50.439078Z", "shell.execute_reply.started": "2021-08-24T18:28:50.315508Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "sns.boxplot(sal[\"years\"], color ='blue')\n", "# No Outlier found based on the boxplot analysis" ] }, { "cell_type": "markdown", "id": "f3fac17e", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "# 3. Apply Simple linear regression with OLS to create different regression model.\n" ] }, { "cell_type": "markdown", "id": "4de2792a", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "## 3.1 Model -1 - Vanilla Model ( No Transformation) - y = ax+b" ] }, { "cell_type": "code", "execution_count": null, "id": "f6c98b87", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:54.135631Z", "iopub.status.busy": "2021-08-24T18:28:54.135174Z", "iopub.status.idle": "2021-08-24T18:28:54.307238Z", "shell.execute_reply": "2021-08-24T18:28:54.306336Z", "shell.execute_reply.started": "2021-08-24T18:28:54.135598Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Check the relation between Variable by Scatter Plot & Correlation Coefficient.\n", "sns.scatterplot(y=\"hike\",x=\"years\",data=sal, color = \"purple\" )\n", "#Relation Type = Linear\n", "#Direction - Positive Correlation\n", "#Strength -Can't Comment\n" ] }, { "cell_type": "code", "execution_count": null, "id": "91d59575", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Find the Correlation Coefficient (R) for the relation\n", "# R tells the magnitude of strength of relation between y & x" ] }, { "cell_type": "code", "execution_count": null, "id": "d8b3b8da", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:58.835548Z", "iopub.status.busy": "2021-08-24T18:28:58.835206Z", "iopub.status.idle": "2021-08-24T18:28:58.843500Z", "shell.execute_reply": "2021-08-24T18:28:58.842326Z", "shell.execute_reply.started": "2021-08-24T18:28:58.835515Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "np.corrcoef(sal[\"years\"],sal[\"hike\"])\n", "# R value is above 0.85 means Correlation strength is High." ] }, { "cell_type": "code", "execution_count": null, "id": "21e210a8", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:00.675497Z", "iopub.status.busy": "2021-08-24T18:29:00.675140Z", "iopub.status.idle": "2021-08-24T18:29:00.776748Z", "shell.execute_reply": "2021-08-24T18:29:00.775787Z", "shell.execute_reply.started": "2021-08-24T18:29:00.675464Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Use OLS & fit model on data\n", "import statsmodels.formula.api as smf\n", "model1= smf.ols('hike~years',data = sal).fit()\n", "model1.summary()" ] }, { "cell_type": "code", "execution_count": null, "id": "ca6874d2", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:03.995742Z", "iopub.status.busy": "2021-08-24T18:29:03.995407Z", "iopub.status.idle": "2021-08-24T18:29:04.283614Z", "shell.execute_reply": "2021-08-24T18:29:04.282947Z", "shell.execute_reply.started": "2021-08-24T18:29:03.995710Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Predict Regression Line Model1\n", "pred1 = model1.predict(pd.DataFrame(sal['years']))\n", "plt.scatter(x=\"years\",y=\"hike\",data=sal, color = \"purple\" )\n", "plt.plot(sal['years'],pred1)\n", "plt.legend(['Predicted line', 'Observed data'])\n", "plt.xlabel('years')\n", "plt.ylabel('Hike')\n", "plt.show()" ] }, { "cell_type": "code", "execution_count": null, "id": "36535505", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:06.500690Z", "iopub.status.busy": "2021-08-24T18:29:06.500185Z", "iopub.status.idle": "2021-08-24T18:29:06.508140Z", "shell.execute_reply": "2021-08-24T18:29:06.507161Z", "shell.execute_reply.started": "2021-08-24T18:29:06.500642Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# RMSE Error calculation (Model-1)\n", "res1 = sal.hike - pred1\n", "res_sqr1 = res1 * res1\n", "mse1 = np.mean(res_sqr1)\n", "rmse1 = np.sqrt(mse1)\n", "rmse1" ] }, { "cell_type": "markdown", "id": "e788491d", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "## 3.2 Model 2 {Log Transformation}, x = log(years); y = hike\n" ] }, { "cell_type": "code", "execution_count": null, "id": "653a3231", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:08.441879Z", "iopub.status.busy": "2021-08-24T18:29:08.441489Z", "iopub.status.idle": "2021-08-24T18:29:08.597219Z", "shell.execute_reply": "2021-08-24T18:29:08.596362Z", "shell.execute_reply.started": "2021-08-24T18:29:08.441847Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Check the relation between Variable by Scatter Plot & Correlation Coefficient.\n", "plt.scatter(x = np.log(sal['years']), y = sal['hike'], color = 'brown')\n", "np.corrcoef(np.log(sal['years']),sal['hike'] ) #correlation\n" ] }, { "cell_type": "code", "execution_count": null, "id": "02855b4e", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:10.356562Z", "iopub.status.busy": "2021-08-24T18:29:10.356189Z", "iopub.status.idle": "2021-08-24T18:29:10.381588Z", "shell.execute_reply": "2021-08-24T18:29:10.380729Z", "shell.execute_reply.started": "2021-08-24T18:29:10.356527Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Use OLS & fit model on data\n", "model2= smf.ols('hike ~ np.log(years)',data = sal).fit()\n", "model2.summary()\n" ] }, { "cell_type": "code", "execution_count": null, "id": "e48bc19e", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:13.498737Z", "iopub.status.busy": "2021-08-24T18:29:13.498399Z", "iopub.status.idle": "2021-08-24T18:29:13.683532Z", "shell.execute_reply": "2021-08-24T18:29:13.682890Z", "shell.execute_reply.started": "2021-08-24T18:29:13.498709Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Predict Regression Line for Model-2\n", "pred2 = model2.predict(pd.DataFrame(sal['years']))\n", "plt.scatter(x=np.log(sal[\"years\"]),y=\"hike\",data= sal, color = \"purple\" )\n", "plt.plot(np.log(sal['years']),pred2)\n", "plt.legend(['Predicted line', 'Observed data'])\n", "plt.xlabel('log(Years Experience)')\n", "plt.ylabel('Salary Hike')\n", "plt.show()" ] }, { "cell_type": "code", "execution_count": null, "id": "230617ee", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:16.276591Z", "iopub.status.busy": "2021-08-24T18:29:16.276135Z", "iopub.status.idle": "2021-08-24T18:29:16.284489Z", "shell.execute_reply": "2021-08-24T18:29:16.283763Z", "shell.execute_reply.started": "2021-08-24T18:29:16.276562Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# RMSE Error calculation for Model-2\n", "res2 = sal.hike- pred2\n", "res_sqr2 = res2 * res2\n", "mse2 = np.mean(res_sqr2)\n", "rmse2 = np.sqrt(mse2)\n", "rmse2" ] }, { "cell_type": "markdown", "id": "6a0d0c66", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "## 3.3 Model-3 {Exponential Method}, y=log(hike), x=year" ] }, { "cell_type": "code", "execution_count": null, "id": "390b5c4e", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:19.276059Z", "iopub.status.busy": "2021-08-24T18:29:19.275549Z", "iopub.status.idle": "2021-08-24T18:29:19.443888Z", "shell.execute_reply": "2021-08-24T18:29:19.442765Z", "shell.execute_reply.started": "2021-08-24T18:29:19.276024Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Check the relation between Variable by Scatter Plot & Correlation Coefficient.\n", "plt.scatter(x = sal['years'], y= np.log(sal['hike']), color = 'brown') # Scatter Plot for checking relation.\n", "np.corrcoef(sal['years'],np.log(sal['hike']) ) #correlation" ] }, { "cell_type": "code", "execution_count": null, "id": "9ea54099", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:22.616519Z", "iopub.status.busy": "2021-08-24T18:29:22.616174Z", "iopub.status.idle": "2021-08-24T18:29:22.640056Z", "shell.execute_reply": "2021-08-24T18:29:22.639023Z", "shell.execute_reply.started": "2021-08-24T18:29:22.616485Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Use OLS & fit model on data\n", "import statsmodels.formula.api as smf\n", "model3= smf.ols('np.log(hike) ~ years',data = sal).fit()\n", "model3.summary()" ] }, { "cell_type": "code", "execution_count": null, "id": "251bb5d7", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:25.630629Z", "iopub.status.busy": "2021-08-24T18:29:25.630270Z", "iopub.status.idle": "2021-08-24T18:29:25.823114Z", "shell.execute_reply": "2021-08-24T18:29:25.822392Z", "shell.execute_reply.started": "2021-08-24T18:29:25.630587Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Predict Regression Line for Model-3\n", "pred3 = model3.predict(pd.DataFrame(sal['years']))\n", "pred3_exp = np.exp(pred3)\n", "plt.scatter(x =(sal['years']), y = np.log(sal['hike']), color = 'brown')\n", "plt.plot(sal['years'],pred3)\n", "plt.legend(['Predicted line', 'Observed data'])\n", "plt.xlabel('Years Experience')\n", "plt.ylabel('log(Salary Hike)')\n", "plt.show()\n" ] }, { "cell_type": "code", "execution_count": null, "id": "54a33425", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:28.176426Z", "iopub.status.busy": "2021-08-24T18:29:28.175928Z", "iopub.status.idle": "2021-08-24T18:29:28.184202Z", "shell.execute_reply": "2021-08-24T18:29:28.183122Z", "shell.execute_reply.started": "2021-08-24T18:29:28.176382Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# RMSE Error calculation\n", "res3 = sal.hike - pred3_exp\n", "res_sqr3 = res3 * res3\n", "mse3 = np.mean(res_sqr3)\n", "rmse3 = np.sqrt(mse3)\n", "rmse3" ] }, { "cell_type": "markdown", "id": "e440d4f2", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "## 3.4 Model-4 Polynomial of Transformation of Degree 2 ( Quadratic) , y=ax^2+bx+c --> {x = years ; x^2 = (years)^2 ; y = hike}" ] }, { "cell_type": "code", "execution_count": null, "id": "e1fd67e2", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:30.355392Z", "iopub.status.busy": "2021-08-24T18:29:30.355061Z", "iopub.status.idle": "2021-08-24T18:29:30.381313Z", "shell.execute_reply": "2021-08-24T18:29:30.380425Z", "shell.execute_reply.started": "2021-08-24T18:29:30.355364Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Use OLS & fit model on data\n", "model4 = smf.ols('hike ~ years + I(yearsyears)', data = sal).fit()\n", "model4.summary()" ] }, { "cell_type": "code", "execution_count": null, "id": "bb571357", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:33.275265Z", "iopub.status.busy": "2021-08-24T18:29:33.274917Z", "iopub.status.idle": "2021-08-24T18:29:33.284717Z", "shell.execute_reply": "2021-08-24T18:29:33.284005Z", "shell.execute_reply.started": "2021-08-24T18:29:33.275237Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "pred4 = model4.predict(pd.DataFrame(sal))\n", "X = sal.iloc[:, 0:1].values\n" ] }, { "cell_type": "code", "execution_count": null, "id": "32b677c2", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:35.375797Z", "iopub.status.busy": "2021-08-24T18:29:35.375247Z", "iopub.status.idle": "2021-08-24T18:29:35.571750Z", "shell.execute_reply": "2021-08-24T18:29:35.571105Z", "shell.execute_reply.started": "2021-08-24T18:29:35.375745Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "plt.scatter(sal.years,sal.hike)\n", "plt.plot(X, pred4, color = 'red')\n", "plt.legend(['Predicted line', 'Observed data'])\n", "plt.show()" ] }, { "cell_type": "code", "execution_count": null, "id": "75b12d63", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:37.760775Z", "iopub.status.busy": "2021-08-24T18:29:37.760219Z", "iopub.status.idle": "2021-08-24T18:29:37.769205Z", "shell.execute_reply": "2021-08-24T18:29:37.768156Z", "shell.execute_reply.started": "2021-08-24T18:29:37.760727Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Error calculation\n", "res4 = sal.hike - pred4\n", "res_sqr4 = res4 res4\n", "mse4 = np.mean(res_sqr4)\n", "rmse4 = np.sqrt(mse4)\n", "rmse4" ] }, { "cell_type": "code", "execution_count": null, "id": "455dd0d5", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:40.715527Z", "iopub.status.busy": "2021-08-24T18:29:40.715157Z", "iopub.status.idle": "2021-08-24T18:29:40.728079Z", "shell.execute_reply": "2021-08-24T18:29:40.727027Z", "shell.execute_reply.started": "2021-08-24T18:29:40.715497Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Choose the best model using RMSE\n", "data = {\"MODEL\":pd.Series([\"SLR\", \"Log model\", \"Exp model\", \"Poly model\"]), \"RMSE\":pd.Series([rmse1, rmse2, rmse3, rmse4])}\n", "table_rmse = pd.DataFrame(data)\n", "table_rmse" ] }, { "cell_type": "code", "execution_count": null, "id": "d6c38340", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Althought the R^2 & RMSE value is higher for Polynomial model but it is having p value higher than 5% Hence we will neglect this model.\n", "#Our Best model is Base Model with No transfomation.( Vanilla Model)" ] }, { "cell_type": "markdown", "id": "6b5fc47e", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "# 4. Best Model Selection from 4 Model tested Based on R , R^2 & RMSE Values" ] }, { "cell_type": "markdown", "id": "f40121eb", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [] }, { "attachments": { "55bb7677-6746-4869-b22f-f446ee366c30.png": { "image/png": "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" } }, "cell_type": "markdown", "id": "3e250a8e", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "![salary.png](attachment:55bb7677-6746-4869-b22f-f446ee366c30.png)" ] }, { "cell_type": "markdown", "id": "f68bfc47", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "## 4.1. BEST MODEL - Vanilla Model ( y=mx+c,No Tranformation)" ] }, { "cell_type": "markdown", "id": "dfa181f5", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "# 5. Train & Test your data on the Best model to check the performance of model on test data" ] }, { "cell_type": "code", "execution_count": null, "id": "107a0dd2", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:31:46.495793Z", "iopub.status.busy": "2021-08-24T18:31:46.495308Z", "iopub.status.idle": "2021-08-24T18:31:46.519385Z", "shell.execute_reply": "2021-08-24T18:31:46.518721Z", "shell.execute_reply.started": "2021-08-24T18:31:46.495751Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "from sklearn.model_selection import train_test_split\n", "\n", "train, test = train_test_split(sal, test_size = 0.25,random_state=6)\n", "\n", "finalmodel = smf.ols('hike ~ years', data = train).fit()\n", "finalmodel.summary()" ] }, { "cell_type": "code", "execution_count": null, "id": "130096fb", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:30:07.301482Z", "iopub.status.busy": "2021-08-24T18:30:07.300762Z", "iopub.status.idle": "2021-08-24T18:30:07.311575Z", "shell.execute_reply": "2021-08-24T18:30:07.310671Z", "shell.execute_reply.started": "2021-08-24T18:30:07.301443Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Prediction on train data\n", "train_pred = finalmodel.predict(pd.DataFrame(train))\n", "train_pred \n" ] }, { "cell_type": "code", "execution_count": null, "id": "484bdf14", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:30:10.175458Z", "iopub.status.busy": "2021-08-24T18:30:10.175124Z", "iopub.status.idle": "2021-08-24T18:30:10.182500Z", "shell.execute_reply": "2021-08-24T18:30:10.181577Z", "shell.execute_reply.started": "2021-08-24T18:30:10.175430Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Model Evaluation on train data\n", "train_res = train.hike - train_pred\n", "train_sqrs = train_res * train_res\n", "train_mse = np.mean(train_sqrs)\n", "train_rmse = np.sqrt(train_mse)\n", "train_rmse" ] }, { "cell_type": "code", "execution_count": null, "id": "4c80bd97", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:30:14.616330Z", "iopub.status.busy": "2021-08-24T18:30:14.615942Z", "iopub.status.idle": "2021-08-24T18:30:14.625987Z", "shell.execute_reply": "2021-08-24T18:30:14.625048Z", "shell.execute_reply.started": "2021-08-24T18:30:14.616302Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Predict on test data\n", "test_pred = finalmodel.predict(pd.DataFrame(test))\n", "test_pred\n" ] }, { "cell_type": "code", "execution_count": null, "id": "9623b027", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:30:18.496575Z", "iopub.status.busy": "2021-08-24T18:30:18.496043Z", "iopub.status.idle": "2021-08-24T18:30:18.504268Z", "shell.execute_reply": "2021-08-24T18:30:18.503206Z", "shell.execute_reply.started": "2021-08-24T18:30:18.496540Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Model Evaluation on Test data\n", "test_res = test.hike - test_pred\n", "test_sqrs = test_res * test_res\n", "test_mse = np.mean(test_sqrs)\n", "test_rmse = np.sqrt(test_mse)\n", "test_rmse" ] }, { "cell_type": "markdown", "id": "41983047", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "# 6 Final Observation\n", "Although Train data is giving better result than test data hence it looks like an overfit model. \n", "But since the RMSE value of Train & Test data is very Close, it can be Inferred Model will perform well in real life scenario with Unknown Data. \n", "Hence, we can predict the Salary Hike with higher accuracy based on Year’s of Experience data.\n", "\n", "\n", "# Final R^2 Value = 97.3%" ] }, { "cell_type": "markdown", "id": "4b4a6793", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.7.10" }, "papermill": { "default_parameters": {}, "duration": 10.836974, "end_time": "2021-08-24T18:33:43.874398", "environment_variables": {}, "exception": true, "input_path": "__notebook__.ipynb", "output_path": "__notebook__.ipynb", "parameters": {}, "start_time": "2021-08-24T18:33:33.037424", "version": "2.3.3" } }, "nbformat": 4, "nbformat_minor": 5 }	0072/954/72954764.ipynb	s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz
{ "cells": [ { "cell_type": "code", "execution_count": 1, "id": "3fa48887", "metadata": { "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19", "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5", "execution": { "iopub.execute_input": "2021-08-24T18:34:01.719783Z", "iopub.status.busy": "2021-08-24T18:34:01.718063Z", "iopub.status.idle": "2021-08-24T18:34:02.991883Z", "shell.execute_reply": "2021-08-24T18:34:02.992444Z", "shell.execute_reply.started": "2021-08-22T07:53:37.022716Z" }, "papermill": { "duration": 1.29219, "end_time": "2021-08-24T18:34:02.992778", "exception": false, "start_time": "2021-08-24T18:34:01.700588", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "import numpy as np\n", "import pandas as pd\n", "from sklearn.model_selection import KFold\n", "from sklearn.metrics import mean_squared_error\n", "from sklearn.preprocessing import OrdinalEncoder\n", "from xgboost import XGBRegressor\n", "import warnings\n", "warnings.filterwarnings('ignore')" ] }, { "cell_type": "code", "execution_count": 2, "id": "f5270513", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:34:03.013045Z", "iopub.status.busy": "2021-08-24T18:34:03.012375Z", "iopub.status.idle": "2021-08-24T18:34:07.911390Z", "shell.execute_reply": "2021-08-24T18:34:07.910351Z", "shell.execute_reply.started": "2021-08-22T07:53:38.097201Z" }, "papermill": { "duration": 4.910101, "end_time": "2021-08-24T18:34:07.911571", "exception": false, "start_time": "2021-08-24T18:34:03.001470", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "train=pd.read_csv('../input/30-days-of-ml/train.csv').drop('id',axis=1)\n", "test=pd.read_csv('../input/30-days-of-ml/test.csv').drop('id',axis=1)\n", "sample=pd.read_csv('../input/30-days-of-ml/sample_submission.csv')" ] }, { "cell_type": "code", "execution_count": 3, "id": "c0056fc5", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:34:07.934236Z", "iopub.status.busy": "2021-08-24T18:34:07.932945Z", "iopub.status.idle": "2021-08-24T18:34:07.936061Z", "shell.execute_reply": "2021-08-24T18:34:07.935533Z", "shell.execute_reply.started": "2021-08-22T07:53:42.319707Z" }, "papermill": { "duration": 0.01685, "end_time": "2021-08-24T18:34:07.936215", "exception": false, "start_time": "2021-08-24T18:34:07.919365", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "object_col=[col for col in train.columns if 'cat' in col]" ] }, { "cell_type": "code", "execution_count": 4, "id": "bac6b141", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:34:07.958424Z", "iopub.status.busy": "2021-08-24T18:34:07.957725Z", "iopub.status.idle": "2021-08-24T18:34:07.962767Z", "shell.execute_reply": "2021-08-24T18:34:07.962094Z", "shell.execute_reply.started": "2021-08-22T07:53:42.325814Z" }, "papermill": { "duration": 0.018904, "end_time": "2021-08-24T18:34:07.962934", "exception": false, "start_time": "2021-08-24T18:34:07.944030", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "xgb_params = {\n", " #'tree_method':'gpu_hist', ## parameters for gpu\n", " #'gpu_id':0, #\n", " #'predictor':'gpu_predictor', #\n", " 'n_estimators': 10000,\n", " 'learning_rate': 0.03628302216953097,\n", " 'subsample': 0.7875490025178415,\n", " 'colsample_bytree': 0.11807135201147481,\n", " 'max_depth': 3,\n", " 'booster': 'gbtree', \n", " 'reg_lambda': 0.0008746338866473539,\n", " 'reg_alpha': 23.13181079976304,\n", " 'n_jobs':-1,\n", " 'random_state':40\n", "}" ] }, { "cell_type": "code", "execution_count": 5, "id": "ddb80e65", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:34:07.991336Z", "iopub.status.busy": "2021-08-24T18:34:07.990423Z", "iopub.status.idle": "2021-08-24T19:17:59.674476Z", "shell.execute_reply": "2021-08-24T19:17:59.675225Z", "shell.execute_reply.started": "2021-08-22T07:54:06.984688Z" }, "papermill": { "duration": 2631.704809, "end_time": "2021-08-24T19:17:59.675635", "exception": false, "start_time": "2021-08-24T18:34:07.970826", "status": "completed" }, "tags": [] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[0]\tvalidation_0-rmse:7.50024\n", "[2000]\tvalidation_0-rmse:0.71913\n", "[4000]\tvalidation_0-rmse:0.71651\n", "[6000]\tvalidation_0-rmse:0.71582\n", "[7940]\tvalidation_0-rmse:0.71567\n", "fold 1 validation error 0.7156685166732855\n", "fold 1 training error 0.705734118918996\n", "--------------------\n", "[0]\tvalidation_0-rmse:7.49703\n", "[2000]\tvalidation_0-rmse:0.71889\n", "[4000]\tvalidation_0-rmse:0.71651\n", "[6000]\tvalidation_0-rmse:0.71597\n", "[6932]\tvalidation_0-rmse:0.71591\n", "fold 2 validation error 0.7159020146190179\n", "fold 2 training error 0.7067970045167419\n", "--------------------\n", "[0]\tvalidation_0-rmse:7.49481\n", "[2000]\tvalidation_0-rmse:0.72070\n", "[4000]\tvalidation_0-rmse:0.71826\n", "[6000]\tvalidation_0-rmse:0.71770\n", "[6506]\tvalidation_0-rmse:0.71769\n", "fold 3 validation error 0.7176803581683315\n", "fold 3 training error 0.7068762736069921\n", "--------------------\n", "[0]\tvalidation_0-rmse:7.49703\n", "[2000]\tvalidation_0-rmse:0.72065\n", "[4000]\tvalidation_0-rmse:0.71818\n", "[6000]\tvalidation_0-rmse:0.71761\n", "[8000]\tvalidation_0-rmse:0.71746\n", "[8115]\tvalidation_0-rmse:0.71747\n", "fold 4 validation error 0.7174377111216177\n", "fold 4 training error 0.705044721965698\n", "--------------------\n", "[0]\tvalidation_0-rmse:7.50253\n", "[2000]\tvalidation_0-rmse:0.71927\n", "[4000]\tvalidation_0-rmse:0.71659\n", "[6000]\tvalidation_0-rmse:0.71593\n", "[8000]\tvalidation_0-rmse:0.71576\n", "[8145]\tvalidation_0-rmse:0.71576\n", "fold 5 validation error 0.7157431615630978\n", "fold 5 training error 0.7055256474596883\n", "--------------------\n" ] } ], "source": [ "model=XGBRegressor(**xgb_params)\n", "i=0\n", "sub=[]\n", "kf=KFold(n_splits=5,shuffle=True,random_state=42)\n", "for train_index,test_index in kf.split(train):\n", " i+=1\n", " xtrain=train.iloc[train_index]\n", " xvalid=train.iloc[test_index]\n", " \n", " df_test=test.copy()\n", " \n", " encoder=OrdinalEncoder()\n", " \n", " xtrain[object_col]=encoder.fit_transform(xtrain[object_col])\n", " xvalid[object_col]=encoder.transform(xvalid[object_col])\n", " df_test[object_col]=encoder.transform(df_test[object_col])\n", " \n", " model.fit(xtrain.drop('target',axis=1),xtrain['target'],\n", " early_stopping_rounds=300,\n", " eval_set=[(xvalid.drop('target',axis=1), xvalid['target'])], \n", " verbose=2000)\n", " \n", " pred_valid=model.predict(xvalid.drop('target',axis=1))\n", " pred_train=model.predict(xtrain.drop('target',axis=1))\n", " print(f'fold {i} validation error ',mean_squared_error(xvalid['target'],pred_valid,squared=False))\n", " print(f'fold {i} training error ',mean_squared_error(xtrain['target'],pred_train,squared=False))\n", " print(\"--------------------\")\n", " \n", " pred=model.predict(df_test)\n", " sub.append(pred)" ] }, { "cell_type": "code", "execution_count": 6, "id": "a2686091", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T19:17:59.731407Z", "iopub.status.busy": "2021-08-24T19:17:59.730490Z", "iopub.status.idle": "2021-08-24T19:17:59.740080Z", "shell.execute_reply": "2021-08-24T19:17:59.739531Z", "shell.execute_reply.started": "2021-08-22T03:56:22.872933Z" }, "papermill": { "duration": 0.033351, "end_time": "2021-08-24T19:17:59.740228", "exception": false, "start_time": "2021-08-24T19:17:59.706877", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "predictions=np.mean(np.column_stack(sub),axis=1)\n", "sample['target']=predictions" ] }, { "cell_type": "code", "execution_count": 7, "id": "c9849b54", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T19:17:59.780142Z", "iopub.status.busy": "2021-08-24T19:17:59.779330Z", "iopub.status.idle": "2021-08-24T19:18:00.353462Z", "shell.execute_reply": "2021-08-24T19:18:00.352789Z", "shell.execute_reply.started": "2021-08-22T03:56:23.969056Z" }, "papermill": { "duration": 0.59623, "end_time": "2021-08-24T19:18:00.353604", "exception": false, "start_time": "2021-08-24T19:17:59.757374", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "sample.to_csv('sub17.csv',index=False)" ] }, { "cell_type": "code", "execution_count": null, "id": "e2444cbb", "metadata": { "papermill": { "duration": 0.016768, "end_time": "2021-08-24T19:18:00.387633", "exception": false, "start_time": "2021-08-24T19:18:00.370865", "status": "completed" }, "tags": [] }, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.7.10" }, "papermill": { "default_parameters": {}, "duration": 2648.326511, "end_time": "2021-08-24T19:18:01.215693", "environment_variables": {}, "exception": null, "input_path": "__notebook__.ipynb", "output_path": "__notebook__.ipynb", "parameters": {}, "start_time": "2021-08-24T18:33:52.889182", "version": "2.3.3" } }, "nbformat": 4, "nbformat_minor": 5 }	0072/954/72954796.ipynb	s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz
"{\"metadata\":{\"kernelspec\":{\"language\":\"python\",\"display_name\":\"Python 3\",\"name\":\"pyt(...TRUNCATED)	0072/954/72954816.ipynb	s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz
"{\n \"cells\": [\n {\n \"cell_type\": \"code\",\n \"execution_count\": 1,\n \"id\": \"93462e(...TRUNCATED)	0072/954/72954919.ipynb	s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz
"{\n \"cells\": [\n {\n \"cell_type\": \"code\",\n \"execution_count\": 1,\n \"id\": \"53716c(...TRUNCATED)	0072/955/72955615.ipynb	s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz
"{\n \"cells\": [\n {\n \"cell_type\": \"code\",\n \"execution_count\": 1,\n \"id\": \"30d1d5(...TRUNCATED)	0072/955/72955804.ipynb	s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz
"{\n \"cells\": [\n {\n \"cell_type\": \"markdown\",\n \"id\": \"c711745b\",\n \"metadata\": (...TRUNCATED)	0072/957/72957076.ipynb	s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz
"{\n \"cells\": [\n {\n \"cell_type\": \"markdown\",\n \"id\": \"3c6fc1f4\",\n \"metadata\": (...TRUNCATED)	0072/957/72957491.ipynb	s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz
"{\"metadata\":{\"kernelspec\":{\"language\":\"python\",\"display_name\":\"Python 3\",\"name\":\"pyt(...TRUNCATED)	0072/957/72957630.ipynb	s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz
"{\n \"cells\": [\n {\n \"cell_type\": \"markdown\",\n \"id\": \"1221ca17\",\n \"metadata\": (...TRUNCATED)	0072/957/72957733.ipynb	s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz

{ "cells": [ { "cell_type": "markdown", "id": "145de7c9", "metadata": { "papermill": { "duration": 0.038294, "end_time": "2021-08-24T18:33:41.181818", "exception": false, "start_time": "2021-08-24T18:33:41.143524", "status": "completed" }, "tags": [] }, "source": [ "# Problem Statement\n", "\n", "The head of HR of a certain organization wants to automate their salary hike estimation. The organization consulted an analytics service provider and asked them to build a basic prediction model by providing them with a dataset that contains the data about the number of years of experience and the salary hike given accordingly. Build a Simple Linear Regression model with salary as the target variable.\n", "\n", "Definition of Variable -:\n", "\n", "1. Year of Experience (Input)\n", "2. Salary Hike -> Hike in salary (Target)\n" ] }, { "cell_type": "markdown", "id": "e5afb14c", "metadata": { "papermill": { "duration": 0.036479, "end_time": "2021-08-24T18:33:41.256896", "exception": false, "start_time": "2021-08-24T18:33:41.220417", "status": "completed" }, "tags": [] }, "source": [ "# Approach to Solve the Business Case -:\n", "1. Business Objective - Find Relationship between Salary Hike and Year of Experience\n", "2. Perform EDA on data (Outlier, Missing Values).\n", "3. Understand the relationship between the variable's using Scatter Plot.\n", "4. Apply Simple linear regression with OLE to create base regression model( Vanilla Model)\n", "5. Check the RMSE, R^2 & R values for the model.\n", "5. Compare the Vanilla model with Model with Transformation's to check the best fit model.\n", " Transformation used\n", " 1. Logarithmic\n", " 2. Exponential\n", " 3. Polynomial with degree 2\n", "6. After comparing RMSE, R^2 & R values, Select the best model.\n", "7. Train & Test your data on these model to check the performance of model on test data.\n", "\n" ] }, { "cell_type": "code", "execution_count": 1, "id": "7c8a58fe", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:33:41.355760Z", "iopub.status.busy": "2021-08-24T18:33:41.345678Z", "iopub.status.idle": "2021-08-24T18:33:42.274231Z", "shell.execute_reply": "2021-08-24T18:33:42.272656Z", "shell.execute_reply.started": "2021-08-24T18:28:29.916029Z" }, "papermill": { "duration": 0.971429, "end_time": "2021-08-24T18:33:42.274538", "exception": true, "start_time": "2021-08-24T18:33:41.303109", "status": "failed" }, "tags": [] }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "UsageError: Line magic function `%` not found.\n" ] } ], "source": [ "#Import Library\n", "import pandas as pd\n", "import numpy as np\n", "import seaborn as sns\n", "import matplotlib.pyplot as plt\n", "% matplotlib inline\n", "import statsmodels.formula.api as smf\n" ] }, { "cell_type": "code", "execution_count": null, "id": "8181f09e", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:31.375993Z", "iopub.status.busy": "2021-08-24T18:28:31.375240Z", "iopub.status.idle": "2021-08-24T18:28:31.387388Z", "shell.execute_reply": "2021-08-24T18:28:31.386508Z", "shell.execute_reply.started": "2021-08-24T18:28:31.375936Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Import the dataset\n", "sal = pd.read_csv(\"../input/salary/Salary.csv\")" ] }, { "cell_type": "code", "execution_count": null, "id": "42db0d3d", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:33.377308Z", "iopub.status.busy": "2021-08-24T18:28:33.375351Z", "iopub.status.idle": "2021-08-24T18:28:33.387130Z", "shell.execute_reply": "2021-08-24T18:28:33.386152Z", "shell.execute_reply.started": "2021-08-24T18:28:33.377271Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Check the data\n", "sal.head()" ] }, { "cell_type": "markdown", "id": "fcd7a320", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "Change the Column Label to short Abbrevation for ease of Code handling.\n", "Weight gained (grams) - wg\n", "Calories Consumed - cc" ] }, { "cell_type": "markdown", "id": "b63802b6", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "### Rename Column for Ease of Coding" ] }, { "cell_type": "code", "execution_count": null, "id": "e326b80b", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:35.875585Z", "iopub.status.busy": "2021-08-24T18:28:35.875250Z", "iopub.status.idle": "2021-08-24T18:28:35.887145Z", "shell.execute_reply": "2021-08-24T18:28:35.886321Z", "shell.execute_reply.started": "2021-08-24T18:28:35.875553Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "sal = sal.rename(columns={\"Salary\":\"hike\",\"YearsExperience\":\"years\"})\n", "sal.head(5)\n" ] }, { "cell_type": "markdown", "id": "73141ddd", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "# 2. Perform EDA on data (Outlier, Missing Values)" ] }, { "cell_type": "markdown", "id": "9edccd42", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "#Exploratory Data Analysis\n", "1. Check for the null Values in dataset.\n", "2. Check for the outlier in dataset\n" ] }, { "cell_type": "code", "execution_count": null, "id": "b3c7ded2", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:39.015306Z", "iopub.status.busy": "2021-08-24T18:28:39.014992Z", "iopub.status.idle": "2021-08-24T18:28:39.029879Z", "shell.execute_reply": "2021-08-24T18:28:39.028852Z", "shell.execute_reply.started": "2021-08-24T18:28:39.015280Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Check for Null Data - No Null values\n", "sal.info()" ] }, { "cell_type": "code", "execution_count": null, "id": "c4369be2", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:41.355559Z", "iopub.status.busy": "2021-08-24T18:28:41.355231Z", "iopub.status.idle": "2021-08-24T18:28:41.373607Z", "shell.execute_reply": "2021-08-24T18:28:41.372617Z", "shell.execute_reply.started": "2021-08-24T18:28:41.355531Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "sal.describe()" ] }, { "cell_type": "code", "execution_count": null, "id": "c99b170a", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:43.395591Z", "iopub.status.busy": "2021-08-24T18:28:43.395282Z", "iopub.status.idle": "2021-08-24T18:28:43.594317Z", "shell.execute_reply": "2021-08-24T18:28:43.593701Z", "shell.execute_reply.started": "2021-08-24T18:28:43.395563Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Plot Histogram to view Distribution of data field ( Univariate)\n", "sns.histplot(sal[\"hike\"], color ='green',kde=True)" ] }, { "cell_type": "code", "execution_count": null, "id": "b8d583c5", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:45.890783Z", "iopub.status.busy": "2021-08-24T18:28:45.890295Z", "iopub.status.idle": "2021-08-24T18:28:46.063523Z", "shell.execute_reply": "2021-08-24T18:28:46.062873Z", "shell.execute_reply.started": "2021-08-24T18:28:45.890751Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Plot Histogram to view Distribution of data field ( Univariate) \n", "sns.histplot(sal[\"years\"], color ='blue',kde=True)" ] }, { "cell_type": "markdown", "id": "7bd0ca60", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "#Outlier Identification\n", "Use BoxPlot to Identify any outlier in data" ] }, { "cell_type": "code", "execution_count": null, "id": "79e7c86b", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:47.955671Z", "iopub.status.busy": "2021-08-24T18:28:47.955171Z", "iopub.status.idle": "2021-08-24T18:28:48.073727Z", "shell.execute_reply": "2021-08-24T18:28:48.072628Z", "shell.execute_reply.started": "2021-08-24T18:28:47.955641Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "sns.boxplot(sal[\"hike\"], color ='green')\n", "# No Outlier found based on the boxplot analysis" ] }, { "cell_type": "code", "execution_count": null, "id": "2a7a75a8", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:50.315537Z", "iopub.status.busy": "2021-08-24T18:28:50.315211Z", "iopub.status.idle": "2021-08-24T18:28:50.439779Z", "shell.execute_reply": "2021-08-24T18:28:50.439078Z", "shell.execute_reply.started": "2021-08-24T18:28:50.315508Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "sns.boxplot(sal[\"years\"], color ='blue')\n", "# No Outlier found based on the boxplot analysis" ] }, { "cell_type": "markdown", "id": "f3fac17e", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "# 3. Apply Simple linear regression with OLS to create different regression model.\n" ] }, { "cell_type": "markdown", "id": "4de2792a", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "## 3.1 Model -1 - Vanilla Model ( No Transformation) - y = ax+b" ] }, { "cell_type": "code", "execution_count": null, "id": "f6c98b87", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:54.135631Z", "iopub.status.busy": "2021-08-24T18:28:54.135174Z", "iopub.status.idle": "2021-08-24T18:28:54.307238Z", "shell.execute_reply": "2021-08-24T18:28:54.306336Z", "shell.execute_reply.started": "2021-08-24T18:28:54.135598Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Check the relation between Variable by Scatter Plot & Correlation Coefficient.\n", "sns.scatterplot(y=\"hike\",x=\"years\",data=sal, color = \"purple\" )\n", "#Relation Type = Linear\n", "#Direction - Positive Correlation\n", "#Strength -Can't Comment\n" ] }, { "cell_type": "code", "execution_count": null, "id": "91d59575", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Find the Correlation Coefficient (R) for the relation\n", "# R tells the magnitude of strength of relation between y & x" ] }, { "cell_type": "code", "execution_count": null, "id": "d8b3b8da", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:28:58.835548Z", "iopub.status.busy": "2021-08-24T18:28:58.835206Z", "iopub.status.idle": "2021-08-24T18:28:58.843500Z", "shell.execute_reply": "2021-08-24T18:28:58.842326Z", "shell.execute_reply.started": "2021-08-24T18:28:58.835515Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "np.corrcoef(sal[\"years\"],sal[\"hike\"])\n", "# R value is above 0.85 means Correlation strength is High." ] }, { "cell_type": "code", "execution_count": null, "id": "21e210a8", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:00.675497Z", "iopub.status.busy": "2021-08-24T18:29:00.675140Z", "iopub.status.idle": "2021-08-24T18:29:00.776748Z", "shell.execute_reply": "2021-08-24T18:29:00.775787Z", "shell.execute_reply.started": "2021-08-24T18:29:00.675464Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Use OLS & fit model on data\n", "import statsmodels.formula.api as smf\n", "model1= smf.ols('hike~years',data = sal).fit()\n", "model1.summary()" ] }, { "cell_type": "code", "execution_count": null, "id": "ca6874d2", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:03.995742Z", "iopub.status.busy": "2021-08-24T18:29:03.995407Z", "iopub.status.idle": "2021-08-24T18:29:04.283614Z", "shell.execute_reply": "2021-08-24T18:29:04.282947Z", "shell.execute_reply.started": "2021-08-24T18:29:03.995710Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Predict Regression Line Model1\n", "pred1 = model1.predict(pd.DataFrame(sal['years']))\n", "plt.scatter(x=\"years\",y=\"hike\",data=sal, color = \"purple\" )\n", "plt.plot(sal['years'],pred1)\n", "plt.legend(['Predicted line', 'Observed data'])\n", "plt.xlabel('years')\n", "plt.ylabel('Hike')\n", "plt.show()" ] }, { "cell_type": "code", "execution_count": null, "id": "36535505", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:06.500690Z", "iopub.status.busy": "2021-08-24T18:29:06.500185Z", "iopub.status.idle": "2021-08-24T18:29:06.508140Z", "shell.execute_reply": "2021-08-24T18:29:06.507161Z", "shell.execute_reply.started": "2021-08-24T18:29:06.500642Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# RMSE Error calculation (Model-1)\n", "res1 = sal.hike - pred1\n", "res_sqr1 = res1 * res1\n", "mse1 = np.mean(res_sqr1)\n", "rmse1 = np.sqrt(mse1)\n", "rmse1" ] }, { "cell_type": "markdown", "id": "e788491d", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "## 3.2 Model 2 {Log Transformation}, x = log(years); y = hike\n" ] }, { "cell_type": "code", "execution_count": null, "id": "653a3231", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:08.441879Z", "iopub.status.busy": "2021-08-24T18:29:08.441489Z", "iopub.status.idle": "2021-08-24T18:29:08.597219Z", "shell.execute_reply": "2021-08-24T18:29:08.596362Z", "shell.execute_reply.started": "2021-08-24T18:29:08.441847Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Check the relation between Variable by Scatter Plot & Correlation Coefficient.\n", "plt.scatter(x = np.log(sal['years']), y = sal['hike'], color = 'brown')\n", "np.corrcoef(np.log(sal['years']),sal['hike'] ) #correlation\n" ] }, { "cell_type": "code", "execution_count": null, "id": "02855b4e", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:10.356562Z", "iopub.status.busy": "2021-08-24T18:29:10.356189Z", "iopub.status.idle": "2021-08-24T18:29:10.381588Z", "shell.execute_reply": "2021-08-24T18:29:10.380729Z", "shell.execute_reply.started": "2021-08-24T18:29:10.356527Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Use OLS & fit model on data\n", "model2= smf.ols('hike ~ np.log(years)',data = sal).fit()\n", "model2.summary()\n" ] }, { "cell_type": "code", "execution_count": null, "id": "e48bc19e", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:13.498737Z", "iopub.status.busy": "2021-08-24T18:29:13.498399Z", "iopub.status.idle": "2021-08-24T18:29:13.683532Z", "shell.execute_reply": "2021-08-24T18:29:13.682890Z", "shell.execute_reply.started": "2021-08-24T18:29:13.498709Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Predict Regression Line for Model-2\n", "pred2 = model2.predict(pd.DataFrame(sal['years']))\n", "plt.scatter(x=np.log(sal[\"years\"]),y=\"hike\",data= sal, color = \"purple\" )\n", "plt.plot(np.log(sal['years']),pred2)\n", "plt.legend(['Predicted line', 'Observed data'])\n", "plt.xlabel('log(Years Experience)')\n", "plt.ylabel('Salary Hike')\n", "plt.show()" ] }, { "cell_type": "code", "execution_count": null, "id": "230617ee", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:16.276591Z", "iopub.status.busy": "2021-08-24T18:29:16.276135Z", "iopub.status.idle": "2021-08-24T18:29:16.284489Z", "shell.execute_reply": "2021-08-24T18:29:16.283763Z", "shell.execute_reply.started": "2021-08-24T18:29:16.276562Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# RMSE Error calculation for Model-2\n", "res2 = sal.hike- pred2\n", "res_sqr2 = res2 * res2\n", "mse2 = np.mean(res_sqr2)\n", "rmse2 = np.sqrt(mse2)\n", "rmse2" ] }, { "cell_type": "markdown", "id": "6a0d0c66", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "## 3.3 Model-3 {Exponential Method}, y=log(hike), x=year" ] }, { "cell_type": "code", "execution_count": null, "id": "390b5c4e", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:19.276059Z", "iopub.status.busy": "2021-08-24T18:29:19.275549Z", "iopub.status.idle": "2021-08-24T18:29:19.443888Z", "shell.execute_reply": "2021-08-24T18:29:19.442765Z", "shell.execute_reply.started": "2021-08-24T18:29:19.276024Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Check the relation between Variable by Scatter Plot & Correlation Coefficient.\n", "plt.scatter(x = sal['years'], y= np.log(sal['hike']), color = 'brown') # Scatter Plot for checking relation.\n", "np.corrcoef(sal['years'],np.log(sal['hike']) ) #correlation" ] }, { "cell_type": "code", "execution_count": null, "id": "9ea54099", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:22.616519Z", "iopub.status.busy": "2021-08-24T18:29:22.616174Z", "iopub.status.idle": "2021-08-24T18:29:22.640056Z", "shell.execute_reply": "2021-08-24T18:29:22.639023Z", "shell.execute_reply.started": "2021-08-24T18:29:22.616485Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Use OLS & fit model on data\n", "import statsmodels.formula.api as smf\n", "model3= smf.ols('np.log(hike) ~ years',data = sal).fit()\n", "model3.summary()" ] }, { "cell_type": "code", "execution_count": null, "id": "251bb5d7", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:25.630629Z", "iopub.status.busy": "2021-08-24T18:29:25.630270Z", "iopub.status.idle": "2021-08-24T18:29:25.823114Z", "shell.execute_reply": "2021-08-24T18:29:25.822392Z", "shell.execute_reply.started": "2021-08-24T18:29:25.630587Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Predict Regression Line for Model-3\n", "pred3 = model3.predict(pd.DataFrame(sal['years']))\n", "pred3_exp = np.exp(pred3)\n", "plt.scatter(x =(sal['years']), y = np.log(sal['hike']), color = 'brown')\n", "plt.plot(sal['years'],pred3)\n", "plt.legend(['Predicted line', 'Observed data'])\n", "plt.xlabel('Years Experience')\n", "plt.ylabel('log(Salary Hike)')\n", "plt.show()\n" ] }, { "cell_type": "code", "execution_count": null, "id": "54a33425", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:28.176426Z", "iopub.status.busy": "2021-08-24T18:29:28.175928Z", "iopub.status.idle": "2021-08-24T18:29:28.184202Z", "shell.execute_reply": "2021-08-24T18:29:28.183122Z", "shell.execute_reply.started": "2021-08-24T18:29:28.176382Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# RMSE Error calculation\n", "res3 = sal.hike - pred3_exp\n", "res_sqr3 = res3 * res3\n", "mse3 = np.mean(res_sqr3)\n", "rmse3 = np.sqrt(mse3)\n", "rmse3" ] }, { "cell_type": "markdown", "id": "e440d4f2", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "## 3.4 Model-4 Polynomial of Transformation of Degree 2 ( Quadratic) , y=ax^2+bx+c --> {x = years ; x^2 = (years)^2 ; y = hike}" ] }, { "cell_type": "code", "execution_count": null, "id": "e1fd67e2", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:30.355392Z", "iopub.status.busy": "2021-08-24T18:29:30.355061Z", "iopub.status.idle": "2021-08-24T18:29:30.381313Z", "shell.execute_reply": "2021-08-24T18:29:30.380425Z", "shell.execute_reply.started": "2021-08-24T18:29:30.355364Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Use OLS & fit model on data\n", "model4 = smf.ols('hike ~ years + I(years*years)', data = sal).fit()\n", "model4.summary()" ] }, { "cell_type": "code", "execution_count": null, "id": "bb571357", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:33.275265Z", "iopub.status.busy": "2021-08-24T18:29:33.274917Z", "iopub.status.idle": "2021-08-24T18:29:33.284717Z", "shell.execute_reply": "2021-08-24T18:29:33.284005Z", "shell.execute_reply.started": "2021-08-24T18:29:33.275237Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "pred4 = model4.predict(pd.DataFrame(sal))\n", "X = sal.iloc[:, 0:1].values\n" ] }, { "cell_type": "code", "execution_count": null, "id": "32b677c2", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:35.375797Z", "iopub.status.busy": "2021-08-24T18:29:35.375247Z", "iopub.status.idle": "2021-08-24T18:29:35.571750Z", "shell.execute_reply": "2021-08-24T18:29:35.571105Z", "shell.execute_reply.started": "2021-08-24T18:29:35.375745Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "plt.scatter(sal.years,sal.hike)\n", "plt.plot(X, pred4, color = 'red')\n", "plt.legend(['Predicted line', 'Observed data'])\n", "plt.show()" ] }, { "cell_type": "code", "execution_count": null, "id": "75b12d63", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:37.760775Z", "iopub.status.busy": "2021-08-24T18:29:37.760219Z", "iopub.status.idle": "2021-08-24T18:29:37.769205Z", "shell.execute_reply": "2021-08-24T18:29:37.768156Z", "shell.execute_reply.started": "2021-08-24T18:29:37.760727Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Error calculation\n", "res4 = sal.hike - pred4\n", "res_sqr4 = res4 * res4\n", "mse4 = np.mean(res_sqr4)\n", "rmse4 = np.sqrt(mse4)\n", "rmse4" ] }, { "cell_type": "code", "execution_count": null, "id": "455dd0d5", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:29:40.715527Z", "iopub.status.busy": "2021-08-24T18:29:40.715157Z", "iopub.status.idle": "2021-08-24T18:29:40.728079Z", "shell.execute_reply": "2021-08-24T18:29:40.727027Z", "shell.execute_reply.started": "2021-08-24T18:29:40.715497Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Choose the best model using RMSE\n", "data = {\"MODEL\":pd.Series([\"SLR\", \"Log model\", \"Exp model\", \"Poly model\"]), \"RMSE\":pd.Series([rmse1, rmse2, rmse3, rmse4])}\n", "table_rmse = pd.DataFrame(data)\n", "table_rmse" ] }, { "cell_type": "code", "execution_count": null, "id": "d6c38340", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "#Althought the R^2 & RMSE value is higher for Polynomial model but it is having p value higher than 5% Hence we will neglect this model.\n", "#Our Best model is Base Model with No transfomation.( Vanilla Model)" ] }, { "cell_type": "markdown", "id": "6b5fc47e", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "# 4. Best Model Selection from 4 Model tested Based on R , R^2 & RMSE Values" ] }, { "cell_type": "markdown", "id": "f40121eb", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [] }, { "attachments": { "55bb7677-6746-4869-b22f-f446ee366c30.png": { "image/png": "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" } }, "cell_type": "markdown", "id": "3e250a8e", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "![salary.png](attachment:55bb7677-6746-4869-b22f-f446ee366c30.png)" ] }, { "cell_type": "markdown", "id": "f68bfc47", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "## 4.1. BEST MODEL - Vanilla Model ( y=mx+c,No Tranformation)" ] }, { "cell_type": "markdown", "id": "dfa181f5", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "# 5. Train & Test your data on the Best model to check the performance of model on test data" ] }, { "cell_type": "code", "execution_count": null, "id": "107a0dd2", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:31:46.495793Z", "iopub.status.busy": "2021-08-24T18:31:46.495308Z", "iopub.status.idle": "2021-08-24T18:31:46.519385Z", "shell.execute_reply": "2021-08-24T18:31:46.518721Z", "shell.execute_reply.started": "2021-08-24T18:31:46.495751Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "from sklearn.model_selection import train_test_split\n", "\n", "train, test = train_test_split(sal, test_size = 0.25,random_state=6)\n", "\n", "finalmodel = smf.ols('hike ~ years', data = train).fit()\n", "finalmodel.summary()" ] }, { "cell_type": "code", "execution_count": null, "id": "130096fb", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:30:07.301482Z", "iopub.status.busy": "2021-08-24T18:30:07.300762Z", "iopub.status.idle": "2021-08-24T18:30:07.311575Z", "shell.execute_reply": "2021-08-24T18:30:07.310671Z", "shell.execute_reply.started": "2021-08-24T18:30:07.301443Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Prediction on train data\n", "train_pred = finalmodel.predict(pd.DataFrame(train))\n", "train_pred \n" ] }, { "cell_type": "code", "execution_count": null, "id": "484bdf14", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:30:10.175458Z", "iopub.status.busy": "2021-08-24T18:30:10.175124Z", "iopub.status.idle": "2021-08-24T18:30:10.182500Z", "shell.execute_reply": "2021-08-24T18:30:10.181577Z", "shell.execute_reply.started": "2021-08-24T18:30:10.175430Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Model Evaluation on train data\n", "train_res = train.hike - train_pred\n", "train_sqrs = train_res * train_res\n", "train_mse = np.mean(train_sqrs)\n", "train_rmse = np.sqrt(train_mse)\n", "train_rmse" ] }, { "cell_type": "code", "execution_count": null, "id": "4c80bd97", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:30:14.616330Z", "iopub.status.busy": "2021-08-24T18:30:14.615942Z", "iopub.status.idle": "2021-08-24T18:30:14.625987Z", "shell.execute_reply": "2021-08-24T18:30:14.625048Z", "shell.execute_reply.started": "2021-08-24T18:30:14.616302Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Predict on test data\n", "test_pred = finalmodel.predict(pd.DataFrame(test))\n", "test_pred\n" ] }, { "cell_type": "code", "execution_count": null, "id": "9623b027", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:30:18.496575Z", "iopub.status.busy": "2021-08-24T18:30:18.496043Z", "iopub.status.idle": "2021-08-24T18:30:18.504268Z", "shell.execute_reply": "2021-08-24T18:30:18.503206Z", "shell.execute_reply.started": "2021-08-24T18:30:18.496540Z" }, "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "outputs": [], "source": [ "# Model Evaluation on Test data\n", "test_res = test.hike - test_pred\n", "test_sqrs = test_res * test_res\n", "test_mse = np.mean(test_sqrs)\n", "test_rmse = np.sqrt(test_mse)\n", "test_rmse" ] }, { "cell_type": "markdown", "id": "41983047", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [ "# 6 Final Observation\n", "Although Train data is giving better result than test data hence it looks like an overfit model. \n", "But since the RMSE value of Train & Test data is very Close, it can be Inferred Model will perform well in real life scenario with Unknown Data. \n", "Hence, we can predict the Salary Hike with higher accuracy based on Year’s of Experience data.\n", "\n", "\n", "# Final R^2 Value = 97.3%" ] }, { "cell_type": "markdown", "id": "4b4a6793", "metadata": { "papermill": { "duration": null, "end_time": null, "exception": null, "start_time": null, "status": "pending" }, "tags": [] }, "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.7.10" }, "papermill": { "default_parameters": {}, "duration": 10.836974, "end_time": "2021-08-24T18:33:43.874398", "environment_variables": {}, "exception": true, "input_path": "__notebook__.ipynb", "output_path": "__notebook__.ipynb", "parameters": {}, "start_time": "2021-08-24T18:33:33.037424", "version": "2.3.3" } }, "nbformat": 4, "nbformat_minor": 5 }

0072/954/72954764.ipynb

s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz

{ "cells": [ { "cell_type": "code", "execution_count": 1, "id": "3fa48887", "metadata": { "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19", "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5", "execution": { "iopub.execute_input": "2021-08-24T18:34:01.719783Z", "iopub.status.busy": "2021-08-24T18:34:01.718063Z", "iopub.status.idle": "2021-08-24T18:34:02.991883Z", "shell.execute_reply": "2021-08-24T18:34:02.992444Z", "shell.execute_reply.started": "2021-08-22T07:53:37.022716Z" }, "papermill": { "duration": 1.29219, "end_time": "2021-08-24T18:34:02.992778", "exception": false, "start_time": "2021-08-24T18:34:01.700588", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "import numpy as np\n", "import pandas as pd\n", "from sklearn.model_selection import KFold\n", "from sklearn.metrics import mean_squared_error\n", "from sklearn.preprocessing import OrdinalEncoder\n", "from xgboost import XGBRegressor\n", "import warnings\n", "warnings.filterwarnings('ignore')" ] }, { "cell_type": "code", "execution_count": 2, "id": "f5270513", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:34:03.013045Z", "iopub.status.busy": "2021-08-24T18:34:03.012375Z", "iopub.status.idle": "2021-08-24T18:34:07.911390Z", "shell.execute_reply": "2021-08-24T18:34:07.910351Z", "shell.execute_reply.started": "2021-08-22T07:53:38.097201Z" }, "papermill": { "duration": 4.910101, "end_time": "2021-08-24T18:34:07.911571", "exception": false, "start_time": "2021-08-24T18:34:03.001470", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "train=pd.read_csv('../input/30-days-of-ml/train.csv').drop('id',axis=1)\n", "test=pd.read_csv('../input/30-days-of-ml/test.csv').drop('id',axis=1)\n", "sample=pd.read_csv('../input/30-days-of-ml/sample_submission.csv')" ] }, { "cell_type": "code", "execution_count": 3, "id": "c0056fc5", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:34:07.934236Z", "iopub.status.busy": "2021-08-24T18:34:07.932945Z", "iopub.status.idle": "2021-08-24T18:34:07.936061Z", "shell.execute_reply": "2021-08-24T18:34:07.935533Z", "shell.execute_reply.started": "2021-08-22T07:53:42.319707Z" }, "papermill": { "duration": 0.01685, "end_time": "2021-08-24T18:34:07.936215", "exception": false, "start_time": "2021-08-24T18:34:07.919365", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "object_col=[col for col in train.columns if 'cat' in col]" ] }, { "cell_type": "code", "execution_count": 4, "id": "bac6b141", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:34:07.958424Z", "iopub.status.busy": "2021-08-24T18:34:07.957725Z", "iopub.status.idle": "2021-08-24T18:34:07.962767Z", "shell.execute_reply": "2021-08-24T18:34:07.962094Z", "shell.execute_reply.started": "2021-08-22T07:53:42.325814Z" }, "papermill": { "duration": 0.018904, "end_time": "2021-08-24T18:34:07.962934", "exception": false, "start_time": "2021-08-24T18:34:07.944030", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "xgb_params = {\n", " #'tree_method':'gpu_hist', ## parameters for gpu\n", " #'gpu_id':0, #\n", " #'predictor':'gpu_predictor', #\n", " 'n_estimators': 10000,\n", " 'learning_rate': 0.03628302216953097,\n", " 'subsample': 0.7875490025178415,\n", " 'colsample_bytree': 0.11807135201147481,\n", " 'max_depth': 3,\n", " 'booster': 'gbtree', \n", " 'reg_lambda': 0.0008746338866473539,\n", " 'reg_alpha': 23.13181079976304,\n", " 'n_jobs':-1,\n", " 'random_state':40\n", "}" ] }, { "cell_type": "code", "execution_count": 5, "id": "ddb80e65", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T18:34:07.991336Z", "iopub.status.busy": "2021-08-24T18:34:07.990423Z", "iopub.status.idle": "2021-08-24T19:17:59.674476Z", "shell.execute_reply": "2021-08-24T19:17:59.675225Z", "shell.execute_reply.started": "2021-08-22T07:54:06.984688Z" }, "papermill": { "duration": 2631.704809, "end_time": "2021-08-24T19:17:59.675635", "exception": false, "start_time": "2021-08-24T18:34:07.970826", "status": "completed" }, "tags": [] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[0]\tvalidation_0-rmse:7.50024\n", "[2000]\tvalidation_0-rmse:0.71913\n", "[4000]\tvalidation_0-rmse:0.71651\n", "[6000]\tvalidation_0-rmse:0.71582\n", "[7940]\tvalidation_0-rmse:0.71567\n", "fold 1 validation error 0.7156685166732855\n", "fold 1 training error 0.705734118918996\n", "--------------------\n", "[0]\tvalidation_0-rmse:7.49703\n", "[2000]\tvalidation_0-rmse:0.71889\n", "[4000]\tvalidation_0-rmse:0.71651\n", "[6000]\tvalidation_0-rmse:0.71597\n", "[6932]\tvalidation_0-rmse:0.71591\n", "fold 2 validation error 0.7159020146190179\n", "fold 2 training error 0.7067970045167419\n", "--------------------\n", "[0]\tvalidation_0-rmse:7.49481\n", "[2000]\tvalidation_0-rmse:0.72070\n", "[4000]\tvalidation_0-rmse:0.71826\n", "[6000]\tvalidation_0-rmse:0.71770\n", "[6506]\tvalidation_0-rmse:0.71769\n", "fold 3 validation error 0.7176803581683315\n", "fold 3 training error 0.7068762736069921\n", "--------------------\n", "[0]\tvalidation_0-rmse:7.49703\n", "[2000]\tvalidation_0-rmse:0.72065\n", "[4000]\tvalidation_0-rmse:0.71818\n", "[6000]\tvalidation_0-rmse:0.71761\n", "[8000]\tvalidation_0-rmse:0.71746\n", "[8115]\tvalidation_0-rmse:0.71747\n", "fold 4 validation error 0.7174377111216177\n", "fold 4 training error 0.705044721965698\n", "--------------------\n", "[0]\tvalidation_0-rmse:7.50253\n", "[2000]\tvalidation_0-rmse:0.71927\n", "[4000]\tvalidation_0-rmse:0.71659\n", "[6000]\tvalidation_0-rmse:0.71593\n", "[8000]\tvalidation_0-rmse:0.71576\n", "[8145]\tvalidation_0-rmse:0.71576\n", "fold 5 validation error 0.7157431615630978\n", "fold 5 training error 0.7055256474596883\n", "--------------------\n" ] } ], "source": [ "model=XGBRegressor(**xgb_params)\n", "i=0\n", "sub=[]\n", "kf=KFold(n_splits=5,shuffle=True,random_state=42)\n", "for train_index,test_index in kf.split(train):\n", " i+=1\n", " xtrain=train.iloc[train_index]\n", " xvalid=train.iloc[test_index]\n", " \n", " df_test=test.copy()\n", " \n", " encoder=OrdinalEncoder()\n", " \n", " xtrain[object_col]=encoder.fit_transform(xtrain[object_col])\n", " xvalid[object_col]=encoder.transform(xvalid[object_col])\n", " df_test[object_col]=encoder.transform(df_test[object_col])\n", " \n", " model.fit(xtrain.drop('target',axis=1),xtrain['target'],\n", " early_stopping_rounds=300,\n", " eval_set=[(xvalid.drop('target',axis=1), xvalid['target'])], \n", " verbose=2000)\n", " \n", " pred_valid=model.predict(xvalid.drop('target',axis=1))\n", " pred_train=model.predict(xtrain.drop('target',axis=1))\n", " print(f'fold {i} validation error ',mean_squared_error(xvalid['target'],pred_valid,squared=False))\n", " print(f'fold {i} training error ',mean_squared_error(xtrain['target'],pred_train,squared=False))\n", " print(\"--------------------\")\n", " \n", " pred=model.predict(df_test)\n", " sub.append(pred)" ] }, { "cell_type": "code", "execution_count": 6, "id": "a2686091", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T19:17:59.731407Z", "iopub.status.busy": "2021-08-24T19:17:59.730490Z", "iopub.status.idle": "2021-08-24T19:17:59.740080Z", "shell.execute_reply": "2021-08-24T19:17:59.739531Z", "shell.execute_reply.started": "2021-08-22T03:56:22.872933Z" }, "papermill": { "duration": 0.033351, "end_time": "2021-08-24T19:17:59.740228", "exception": false, "start_time": "2021-08-24T19:17:59.706877", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "predictions=np.mean(np.column_stack(sub),axis=1)\n", "sample['target']=predictions" ] }, { "cell_type": "code", "execution_count": 7, "id": "c9849b54", "metadata": { "execution": { "iopub.execute_input": "2021-08-24T19:17:59.780142Z", "iopub.status.busy": "2021-08-24T19:17:59.779330Z", "iopub.status.idle": "2021-08-24T19:18:00.353462Z", "shell.execute_reply": "2021-08-24T19:18:00.352789Z", "shell.execute_reply.started": "2021-08-22T03:56:23.969056Z" }, "papermill": { "duration": 0.59623, "end_time": "2021-08-24T19:18:00.353604", "exception": false, "start_time": "2021-08-24T19:17:59.757374", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "sample.to_csv('sub17.csv',index=False)" ] }, { "cell_type": "code", "execution_count": null, "id": "e2444cbb", "metadata": { "papermill": { "duration": 0.016768, "end_time": "2021-08-24T19:18:00.387633", "exception": false, "start_time": "2021-08-24T19:18:00.370865", "status": "completed" }, "tags": [] }, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.7.10" }, "papermill": { "default_parameters": {}, "duration": 2648.326511, "end_time": "2021-08-24T19:18:01.215693", "environment_variables": {}, "exception": null, "input_path": "__notebook__.ipynb", "output_path": "__notebook__.ipynb", "parameters": {}, "start_time": "2021-08-24T18:33:52.889182", "version": "2.3.3" } }, "nbformat": 4, "nbformat_minor": 5 }

0072/954/72954796.ipynb

s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz

"{\"metadata\":{\"kernelspec\":{\"language\":\"python\",\"display_name\":\"Python 3\",\"name\":\"pyt(...TRUNCATED)

0072/954/72954816.ipynb

s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz

"{\n \"cells\": [\n {\n \"cell_type\": \"code\",\n \"execution_count\": 1,\n \"id\": \"93462e(...TRUNCATED)

0072/954/72954919.ipynb

s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz

"{\n \"cells\": [\n {\n \"cell_type\": \"code\",\n \"execution_count\": 1,\n \"id\": \"53716c(...TRUNCATED)

0072/955/72955615.ipynb

s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz

"{\n \"cells\": [\n {\n \"cell_type\": \"code\",\n \"execution_count\": 1,\n \"id\": \"30d1d5(...TRUNCATED)

0072/955/72955804.ipynb

s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz

"{\n \"cells\": [\n {\n \"cell_type\": \"markdown\",\n \"id\": \"c711745b\",\n \"metadata\": (...TRUNCATED)

0072/957/72957076.ipynb

s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz

"{\n \"cells\": [\n {\n \"cell_type\": \"markdown\",\n \"id\": \"3c6fc1f4\",\n \"metadata\": (...TRUNCATED)

0072/957/72957491.ipynb

s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz

"{\"metadata\":{\"kernelspec\":{\"language\":\"python\",\"display_name\":\"Python 3\",\"name\":\"pyt(...TRUNCATED)

0072/957/72957630.ipynb

s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz

"{\n \"cells\": [\n {\n \"cell_type\": \"markdown\",\n \"id\": \"1221ca17\",\n \"metadata\": (...TRUNCATED)

0072/957/72957733.ipynb

s3://data-agents/kaggle-outputs/sharded/025_00072.jsonl.gz