Hello,
I am a beginner with Knime.
After I was able to use AutoML for classification without major problems, I wanted to use AutoML (regression) to predict sales up to 31 December 2025.
Principally the workflow works, but there are a few problems and questions.
Regarding the data, I have a data set with sales data since 2023 with 8 SKUs and daily stock levels (1 = low stock, 0 = sufficient stock).
-
in my data there are days especially Saturdays and Sundays without sales, in the forecast every day has the same sales per SKU. Although I would expect a change depending on the day.
-
since AutoML (regression) can only predict one column, I have decided in favour of sales. Is it possible to make a forecast for several items, or do I have to forecast the sales first and then the stock accordingly?
-
to get a table with the forecast, I upload an excel sheet without sales data, so only with date and SKU and then the forecast is made in the Workflok excuter. Is there an easy way to do this so that I don’t have to create an Excel with one entry per SKU, per day.
Many thanks in advance and if there is a course/book that deals more intensively with time series analysis, I will be pleased to receive any information.