For example, i have sales data by categories | Date | GE | VIC | | -- | -- | --| |03.01.2022 |2|7| |10.01.2022 |30 |12| |17.01.2022 |15 |5| |24.01.2022 |57 |8| |.....|...|...| |28.08.2023 |16 |2| And I’m trying to create a forecast for the next 3 months. I’m using ARIMA for it but also tried LSTM and Prophet. My boss like ARIMA because you can try to explain ARIMA parameters with real world things like demand cycles.

My overall forecasting results are not so good. MAPE for some categories more than 80%. Only a few categories have MAPE less than 20%.

At the same time, I also have warehouse stocks data for the same categories for the same dates.

Date GE GE (without replenishments)
03.01.2022 730 730
10.01.2022 700 694
17.01.2022 685 677
..... ... ...
28.08.2023 580 -3198

Can I use that data for forecasting? If yes, how i should convert it for sales? Should i use leftovers data without replenishments?

I will also need to forecast items directly, without grouping categories. And the majority of items don’t have regular sales but have regular stocks data. So, I’m trying to figure out will it be helpful in any way. I have access to our sales system, so, I can collect any data that stored there but don’t think that there may be anything interesting besides sales and stocks data. Thanks.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.