---

BigMart Sales Prediction using PCA and Decision Trees

This project applies Principal Component Analysis (PCA) to the BigMart dataset, followed by a decision tree model to predict sales.

Overview

The analysis uses PCA for dimensionality reduction on the BigMart dataset, then builds a decision tree model to predict Item_Outlet_Sales. This approach helps in understanding the underlying structure of the data and potentially improves prediction accuracy.

Key Features

• Data preprocessing and handling of missing values
• PCA for dimensionality reduction
• Decision tree model for sales prediction
• Visualization of PCA results and decision tree

Requirements

• R (version used in your environment)
• Required packages: data.table, tidyverse, fastDummies, rpart, rattle

Usage

1. Set your working directory in the script
2. Ensure all required CSV files are in the working directory
3. Run the R script

Data Preprocessing

• Handling of missing values in Item_Weight and Item_Visibility
• Creation of dummy variables for categorical features
• Merging of train and test datasets for consistent preprocessing

PCA Analysis

• PCA applied to the preprocessed dataset
• Scree plot and cumulative variance plot generated for PC selection

Machine Learning

• Decision tree model built using the rpart package
• Prediction on test data using the trained model

Results

The final predictions are saved in "Predictions.csv".

---