Hey everyone,

I’m currently building a time-series forecasting model for retail demand, and I’ve hit a bit of a roadblock when it comes to selecting the right seasonal parameters. The data clearly shows seasonal patterns, but I’m unsure about how to determine the most effective seasonality configuration, especially when using models like SARIMA or Prophet.

Specifically, I’m hoping to get advice on:

Seasonality detection: What’s the best way to identify the correct seasonal cycle length in data that shows multiple repeating patterns—weekly, monthly, or even yearly? Model tuning: How do I fine-tune parameters like seasonal order in SARIMA or seasonality modes in Prophet to best reflect the underlying patterns? Evaluation strategies: What are the best practices to validate whether the chosen seasonal setup improves model performance, and how do I avoid overfitting?

If you’ve worked with seasonality in time-series data before, I’d love to hear how you approached it and what worked for you. Thanks in advance!

Best, Survival Race