Time-Series-Sales-Forecasting

Time Series Sales Forecasting: A project that leverages time series and causal models such as ARIMA, Holt-Winters, and Linear Regression to forecast product sales and identify trends for data-driven decision-making.

Project Objective

The goal of this project was to apply various quantitative methods (i.e., Time Series Models and Causal Models) to forecast product sales using historical data. The project involved:

• Performing time series analysis to identify patterns and trends in sales data.
• Applying multiple forecasting models on the training dataset.
• Selecting the best-performing model to forecast sales on the test data.

Models Covered:

1. Seasonal Naive Model
2. Holt-Winters Model (Triple Exponential Smoothing)
3. ARIMA and Seasonal ARIMA Models
4. Linear Regression Model

Performance Metrics:

The following metrics were used to evaluate and compare the performance of each forecasting model:

• Mean Absolute Error (MAE): Measures the average magnitude of the errors in a set of predictions.
• Mean Squared Error (MSE): Indicates the average squared difference between observed and predicted values.
• Root Mean Squared Error (RMSE): Provides a measure of the standard deviation of the prediction errors.
• Mean Absolute Percentage Error (MAPE): Assesses the accuracy of forecasts as a percentage.

By leveraging these models and metrics, the project aimed to improve forecasting accuracy and provide valuable insights into sales trends and future demand.

Source Code:

You can access the complete code and dataset on GitHub: Sales Forecasting Repository