Manufacturing Demand Forecasting

Manufacturing Demand Forecasting

This workflow performs time series forecasting on automotive product demand for a manufacturing company using SARIMA models. The dataset contains historical data (2020-2024) on weekly produced quantities for three different products (winter tires, A/C compressors, and windshield wipers), the date stamp, as well as additional manufacturing information (e.g., the production value, scrap rate, plant temperature, etc.). The goal is to help the manufacturer accurately forecast the quantities to produce in order to avoid out-of-stock events and empty shelves for customers.

• The workflow accesses the dataset, allows product selection, aggregates values to weekly granularity, and visually explores the time series.

• Splits data into training (2020–2023) and testing (2024), trains SARIMA models, and generates forecasts.

• Finally, it joins results, extends forecasts into 2025–2026, and visualizes/scores performance with metrics and plots.

Data Reading

Data Preparation and Time Series Exploration

Model Training and Forecast Generation

Visualize and Score Forecasts

• Weekly quantity forecasts for 2024, using 2020-2023 training data

• Weekly quantity forecasts for entire time series, using 2020-2024 training data

• Weekly quantity forecasts for future years (2025-2026)