Bike Store Sales Forecasting

Overview

This repository contains the code and dataset for forecasting the order quantity of a bike store chain's products based on historical sales data from 2011 to 2015. Developed a Gradient Boosting Regression model and a Random Forest Regressor model to predict future order quantities.

Table of Contents

• Dataset
• Notebook
• Results

Dataset

The dataset used for this project is available in Sales.csv. It includes historical sales data from 2011 to 2015, which serves as the foundation for training and testing the predictive model.

Notebook

bike_sales_forecasting.ipynb contains is the Jupyter Notebook for this project. You can explore the entire data preprocessing, feature engineering, model development, and evaluation process in this notebook.

Results

Here are the final evaluation metrics for both the models

For RandomForestRegressor :

Data	RMSE	r2_score
Train	4.04	82.34%
Val	4.72	74.50%
Test	4.77	73.98%

For GradientBoostingRegressor:

Data	RMSE	r2_score
Train	3.76	84.70%
Val	4.16	80.18%
Test	4.24	79.50%