I have weekly sales data for different products. After taking into account several effects, I used a Seasonal-Trend decomposition (STL) to extract the seasonality of the time series. In a next step, I want to quantify the significant/magnitude of the seasonality for each product and each week. In a first step, I calculated the strength of the seasonality for each product following Rob Hyndman's suggestio here: https://otexts.com/fpp2/seasonal-strength.html. This gives me kind of a global metric for each product but not on a weekly basis.
One idea I had, is to simply normalize all the values of the seasonality (for each product) to be between 0 and 1. However, if the product is not seasonal at all, this value is misleading. Any idea or though is highly appreciated.
