172 lines (171 with data), 8.0 kB

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd  \n",
    "import matplotlib.pylab as plt \n",
    "from matplotlib import pyplot as plt1\n",
    "import seaborn as sns \n",
    "from sklearn.model_selection import train_test_split \n",
    "# read the datafile using panda library.  ensure right file location on machine. \n",
    "data = pd.read_csv(r\"C:\\Users\\SAARTH CHAHAL\\Desktop\\Programming\\AIML\\smoking_driking_dataset_Ver01.csv\")\n",
    "# EDA (Exploratory Data Analysis): \n",
    "# Determine number of rows and colums in the provided data\n",
    "data.shape \n",
    "# print(data.shape)\n",
    "data.head()\n",
    "data.columns\n",
    "data.nunique(axis=0)\n",
    "\n",
    "# Describe to Understanding the dataset. \n",
    "data.describe().apply(lambda s: s.apply(lambda x: format(x, 'f')))\n",
    "\n",
    "# Cleaning dataset by removing null values using method.  on visual examination of data there was no null value\n",
    "data_cleaned = data.dropna(axis=0)\n",
    "print(data_cleaned.shape) \n",
    "\n",
    "# Cleaning  dataset by removing outliers \n",
    "# waistline range of 25 to 150 is based on observation of the data \n",
    "data_cleaned = data_cleaned[data_cleaned['waistline'].between(25,150)]   \n",
    "# sight_left above 5 is based on observation of the data  \n",
    "data_cleaned = data_cleaned[data_cleaned['sight_left'] < 5 ]\n",
    "# sight_right above 5 is based on observation of the data  \n",
    "data_cleaned = data_cleaned[data_cleaned['sight_right'] < 5 ]\n",
    "#since in correlation down the line we will require all number we will need to drop sex which takes string as input.  \n",
    "data_cleaned = data_cleaned.drop('sex',axis=1) \n",
    "# convert drinker as Y or N \n",
    "data_cleaned['DRK_YN'] = np.where(data_cleaned['DRK_YN'] == 'Y', 1,0 ) \n",
    "\n",
    "\n",
    "data_cleaned.shape\n",
    "\n",
    "data_cleaned.describe().apply(lambda s: s.apply(lambda x: format(x, 'f')))\n",
    "\n",
    "# Able to reduce 5803 records which are outliers in the data. \n",
    "# not changing any data related to BP as data seems to be in range. \n",
    "\n",
    "# Data Plotting exercise\n",
    "# to analyze relation ship between variables. \n",
    "# calculate the correlation matrix.   \n",
    "# There are too many variables to produce more readable correlation matrix and heatmap\n",
    "# Created 2 smaller array for matrix and heatmap for smoke and drink correlation \n",
    "\n",
    "dfdata= pd.DataFrame(data_cleaned) \n",
    "dfdata_smk=dfdata[['tot_chole','HDL_chole','LDL_chole','triglyceride','hemoglobin','SMK_stat_type_cd']]\n",
    "dfdata_drk=dfdata[['urine_protein','serum_creatinine','SGOT_AST','SGOT_ALT','gamma_GTP','DRK_YN']]\n",
    "corr_matrix_smk = dfdata_smk.corr()\n",
    "corr_matrix_drk = dfdata_drk.corr()\n",
    "\n",
    "# plot the heatmap \n",
    "sns.heatmap(corr_matrix_smk, xticklabels=corr_matrix_smk.columns, yticklabels=corr_matrix_smk.columns, annot=True, cmap=sns.diverging_palette(220, 20, as_cmap=True))\n",
    "sns.heatmap(corr_matrix_drk, xticklabels=corr_matrix_drk.columns, yticklabels=corr_matrix_drk.columns, annot=True, cmap=sns.diverging_palette(220, 20, as_cmap=True))\n",
    "\n",
    "# scatter plots for two variables \n",
    "dfdata.plot(kind='scatter', x='tot_chole', y='SMK_stat_type_cd')\n",
    "dfdata.plot(kind='scatter', x='SGOT_AST', y='DRK_YN') \n",
    "\n",
    "# sns.pairplot for few variables \n",
    "sns.pairplot ( dfdata ,\n",
    "x_vars=[\"age\" ,\"waistline\", \"tot_chole\" ,  \"SGOT_AST\"  , \"SMK_stat_type_cd\"   , \"DRK_YN\" ], \n",
    "y_vars=[\"age\" ,\"waistline\", \"tot_chole\" ,  \"SGOT_AST\"] , ) \n",
    "\n",
    "# Model training Module \n",
    "# Learning model \n",
    "\n",
    "from sklearn.model_selection import train_test_split \n",
    "# Train learning regression model  \n",
    "# We will need to first split up our data into an X1 array(cholesterol)  that contains the features to train on, \n",
    "# And a y1 array(SMK_stat_type_cd) with the target variable, \n",
    "X1=dfdata[['tot_chole','HDL_chole','LDL_chole','triglyceride','hemoglobin']]\n",
    "y1=dfdata['SMK_stat_type_cd']\n",
    "# split up our data into an X2 array(Kidney function) that contains the features to train on, \n",
    "# And a y2 array(DRK_YN)\n",
    "X2=dfdata[['urine_protein','serum_creatinine','SGOT_AST','SGOT_ALT','gamma_GTP']]\n",
    "y2=dfdata['DRK_YN']\n",
    "\n",
    "# Train test split. test split is 40 % train set is 60 % \n",
    "X1_train, X1_test, y1_train, y1_test = train_test_split(X1, y1, test_size=0.4, random_state=42)\n",
    "X2_train, X2_test, y2_train, y2_test = train_test_split(X2, y2, test_size=0.4, random_state=42)\n",
    "\n",
    "# #Loading the linear regression Model\n",
    "\n",
    "from sklearn.linear_model import LinearRegression \n",
    "\n",
    "lm1 = LinearRegression() \n",
    "lm2= LinearRegression() \n",
    "lm1.fit(X1_train,y1_train)  \n",
    "lm2.fit(X2_train,y2_train) \n",
    "# prediction on Training data  \n",
    "# prediction on Training data \n",
    "training_data_prediction1 = lm1.predict(X1_train) \n",
    "training_data_prediction2 = lm2.predict(X2_train) \n",
    "\n",
    "# Model evlauation.  \n",
    "# Let's evaluate the model by checking out it's coefficients and how we can interpret them.\n",
    "print(lm1.intercept_)\n",
    "coeff_df1 = pd.DataFrame(lm1.coef_,X1.columns,columns=['Coefficient'])\n",
    "coeff_df1 \n",
    "print(lm2.intercept_)\n",
    "coeff_df2 = pd.DataFrame(lm2.coef_,X2.columns,columns=['Coefficient'])\n",
    "coeff_df2 \n",
    "## interpreting the coefficient.\n",
    "# For every one unit change in smoke status there is negative impact on Cholestrol ( refelcted as negative)\n",
    "# and increase in  triglyceride and  hemoglobin which negatively affect the health indicator. \n",
    "\n",
    "# # Prediction from Model. \n",
    "\n",
    "predictions = lm1.predict(X1_test)\n",
    "predictions = lm2.predict(X2_test)\n",
    "plt1.scatter(y1_test,predictions)\n",
    "sns.displot((y1_test-predictions),bins=50); \n",
    "plt1.scatter(y2_test,predictions)\n",
    "sns.displot((y2_test-predictions),bins=50);\n",
    "\n",
    "# Regression Evaluation Metrics\n",
    "# Here are three common evaluation metrics for regression problems:\n",
    "# Mean Absolute Error** (MAE) is the mean of the absolute value of the errors: is the easiest to understand, because it's the average error.\n",
    "# Mean Squared Error** (MSE) is the mean of the squared errors: is more popular than MAE, because MSE \"punishes\" larger errors, which tends to be useful in the real world.\n",
    "# Root Mean Squared Error** (RMSE) is the square root of the mean of the squared errors: is even more popular than MSE, because RMSE is interpretable in the \"y\" units.\n",
    "\n",
    "from sklearn import metrics\n",
    "print('MAE:1',metrics.mean_absolute_error(y1_test, predictions))\n",
    "print('MSE:1',metrics.mean_squared_error(y1_test, predictions))\n",
    "print('RMSE:1',np.sqrt(metrics.mean_squared_error(y1_test, predictions)))\n",
    "\n",
    "print('MAE:2',metrics.mean_absolute_error(y2_test, predictions))\n",
    "print('MSE:2',metrics.mean_squared_error(y2_test, predictions))\n",
    "print('RMSE:2',np.sqrt(metrics.mean_squared_error(y2_test, predictions)))\n",
    "\n",
    "\n",
    "\n"
   ]
  }
 ],
 "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": "3.11.5"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}