Simple Linear Regression Evaluation

This project contains a Jupyter Notebook that explores a marketing and sales dataset, identifies the strongest advertising channel for predicting sales, and fits a simple linear regression model using Ordinary Least Squares (OLS).

The analysis is implemented in Linear_Regression_analysis.ipynb and includes data inspection, missing-value handling, exploratory visualizations, correlation analysis, model fitting, regression diagnostics, and a plain-language interpretation of the results.

Project Overview

• Loads the marketing_and_sales_data_evaluate_lr.csv dataset
• Reviews the dataset structure, data types, and summary statistics
• Fills missing numeric values with the median
• Generates exploratory plots for variable distributions and channel-vs-sales relationships
• Builds a correlation heatmap to compare predictors
• Selects the strongest single predictor for sales
• Fits an OLS simple linear regression model
• Produces regression diagnostics and an interpretation summary

Key Findings

• Dataset size: 4,572 rows and 4 columns
• Variables: TV, Radio, Social_Media, Sales
• Best predictor of Sales: TV
• Correlation with Sales:
  • TV: 0.9966
  • Radio: 0.8674
  • Social_Media: 0.5281
• Final model formula: Sales ~ TV
• Model performance:
  • R-squared: 0.9933
  • Coefficient for TV: 3.5545
  • Intercept: 0.2923
  • P-value: 0.000000

Interpretation: for every $1 increase in TV advertising budget, predicted sales increase by about $3.55, based on this fitted simple linear regression model.

Project Files

• Linear_Regression_analysis.ipynb - main notebook for the full analysis
• marketing_and_sales_data_evaluate_lr.csv - dataset used in the notebook
• plot_distributions.png - variable distribution plots
• plot_scatter.png - scatter plots of marketing channels against sales
• plot_correlation_heatmap.png - feature correlation heatmap
• plot_regression_line.png - fitted regression line plot
• plot_diagnostics.png - regression diagnostic plots

Environment Setup

1. Create a virtual environment

python -m venv venv

2. Activate the virtual environment

On Windows:

venv\Scripts\activate

On macOS/Linux:

source venv/bin/activate

3. Install required packages

pip install notebook pandas numpy matplotlib seaborn statsmodels scipy

4. Launch Jupyter Notebook

jupyter notebook

Then open Linear_Regression_analysis.ipynb.

How To Run

1. Activate your virtual environment.
2. Launch Jupyter Notebook.
3. Open Linear_Regression_analysis.ipynb.
4. Run the notebook cells from top to bottom.

Notes

• The notebook imputes missing values using the median for each numeric column.
• Diagnostic plots are included to check linearity, residual normality, and homoscedasticity.
• This project focuses on simple linear regression using a single best predictor rather than a multiple regression model.