Hi! I’m working on a demand forecasting problem with 44,000 unique IDs, and I have 1.5 years of historical data, resulting in a total of 30 million rows. Most of the data exhibits strong weekly seasonality (every 7 days). I’m considering using Nixtla for this task. Given the size and nature of the dataset, can I approach this using any method from Nixtla’s library, or are there specific methods or models that are better suited for handling datasets with such characteristics (e.g., high cardinality, strong seasonality)? My main priority is achieving a good RMSE while ensuring scalability of the solution. Would love to hear your recommendations!

Hi, Yes, you can use any of our packages. Be aware that StatsForecast models typically are trained on a single series, so such a solution may be slow for this type of problem. Demand forecasting for this type of problem is really common. I'd personally start with a StatsForecast baseline (SeasonalNaive), then create a LightGBM-type model, and maybe try a neural network. Iterate from there 🙂