This Python script performs an end-to-end data analysis and prediction process on an e-commerce dataset using machine learning techniques.

1. Load & Clean Data:

   • Reads data from ecommerce_data.csv.
   • Converts columns to appropriate data types.
   • Fills missing values with column means (excluding CustomerID).

2. Exploratory Data Analysis (EDA):

   • Generates a pair plot and a heatmap to visualize correlations.

3. Feature Selection & Train-Test Split:

   • Selects key features (Customer_Age, Customer_Satisfaction, Number_of_Purchases) to predict Revenue.
   • Splits data into training and testing sets.

4. Model Training & Evaluation:

   • Compares RandomForestRegressor and XGBoostRegressor using cross-validation.
   • Chooses the best model based on R² score.
   • Performs hyperparameter tuning using GridSearchCV if XGBoost is the best model.

5. Final Evaluation:

   • Computes Mean Absolute Error (MAE) and R² score for the best model.
   • Generates a stack plot comparing actual vs predicted revenue.

6. Data Visualization:

   • Stem plots for customer age vs revenue and satisfaction.
   • Stack plots to visualize relationships between variables.
   • Pie chart for revenue distribution.
   • Bar plots for customer satisfaction and number of purchases.

This script effectively cleans data, trains machine learning models, evaluates performance, and visualizes important insights for decision-making in e-commerce analytics.