Forecasting

Overview

Welcome to our introductory course on sales forecasting using machine learning. In this course, we focus on predicting future sales based on historical data using a time‑series model called SARIMA (Seasonal AutoRegressive Integrated Moving Average).

SARIMA is well‑suited to this task because it captures both trend and seasonal patterns in time‑series data, which are common in sales. It extends the ARIMA model by adding seasonal components, making it ideal for data with recurring patterns.

While SARIMA is powerful, it’s not the only forecasting method. Other approaches include exponential smoothing, Facebook’s Prophet, and machine‑learning models such as LSTM neural networks. SARIMA is especially useful when you want a statistical model that’s interpretable and can represent seasonality explicitly.

We’ll start by visualizing the sales data, identifying trends and seasonality, then build a SARIMA model to generate forecasts.