Simple Linear Regression

Sales Prediction on Advertisement Spend

This project leverages Simple Linear Regression to model and predict product sales based on advertisement expenditures across different media channels. Using a dataset sourced from Kaggle, the analysis identifies relationships between advertisement spending (TV, Radio, and Newspaper) and product sales. The primary focus of the project is to build a predictive model that effectively forecasts sales based on TV advertisement spend.

Github

Project Objectives

Understand the correlation between advertisement spending and sales.
Develop a regression model to predict sales using TV ad spend as the independent variable.
Evaluate the performance of the model using statistical metrics and provide insights through visualization.

Dataset Details

Source: Kaggle Advertising Dataset
Key Features:
- Independent Variables: TV, Radio, Newspaper Advertisement Spend
- Target Variable: Sales
Dataset Summary:
- Records: 200
- Variables: 3 Features + 1 Target (Sales)

Key Methodology

Step-1: Data Loading and Cleaning:
- Loaded the dataset from Kaggle.
- Conducted initial data checks to ensure no missing or null values.
- Standardized features for uniform scaling using StandardScaler.
Step-2 : Exploratory Data Analysis (EDA):
- Analyzed correlations using a correlation matrix.
- Visualized data distributions and relationships through scatter plots and pair plots.
Step-3 : Data Splitting:
- Split the dataset into Training (75%) and Testing (25%) sets using train_test_split.
Step-4 : Model Development:
- Trained a Simple Linear Regression Model with TV Advertisement Spend as the predictor for Sales using scikit-learn's LinearRegression.
Step-5 : Performance Evaluation:
- Evaluated the model using:
  - MSE (Mean Squared Error): 5.51
  - MAE (Mean Absolute Error): 1.87
  - RMSE (Root Mean Squared Error): 2.35
  - R² Score: 0.797
  - Adjusted R² Score: 0.793
- Compared actual v/s predicted sales on the test set.
- Visualization:
  - Plotted actual v/s predicted sales on a scatter plot for validation.

Key Insights

Strong Correlation:
- TV advertisement spend shows a high correlation (0.90) with sales, making it an effective predictor.
Model Performance::
- The model explains ~80% of the variance in sales, showcasing its effectiveness as a linear predictor.
Actionable Inight:
- Increased spending on TV advertisements is positively correlated with higher sales.

Code Workflow

Libraries Used

- Pandas
- NumPy
- Matplotlib
- Seaborn
- Scikit-learn

Model Training

from sklearn.linear_model import LinearRegression

model = LinearRegression()

model.fit(X_train, y_train)

Performance Metrics

from sklearn.metrics import mean_squared_error, mean_absolute_error, r2_score

mse = mean_squared_error(y_test, y_pred)

mae = mean_absolute_error(y_test, y_pred)

rmse = np.sqrt(mse)

r2 = r2_score(y_test, y_pred)

Conclusion

This project demonstrates the predictive power of simple linear regression in understanding the relationship between advertisement spending and sales. The findings indicate that TV advertisement spending has a strong positive correlation (0.90) with sales, making it a significant factor in driving revenue.

The regression model effectively explains approximately 80% of the variance in sales, as reflected by the R² score of 0.797, validating its utility as a forecasting tool

For more details, check my github repository: Click Here

Page updated

Google Sites

Report abuse