I need to make sales forecast for next 16 months (last 4 months in 2019 and 12 months in 2020) and I have past 5 years (4x12months+8 months) sales forecast in monthly based (8 months in 2019).
so far, I have been using Minitab -> Winter’s Hold method. However, in order to make forecast in Knime which node should I use? ARIMA, linear regression etc.
thanks for advancesales forecast.xlsx (12.7 KB)
What type of forecast you use is extremely dependent on the business problem you’re trying to solve, the nature of the businesses seasonality, and who will be using/interpreting the model. When I did demand forecasting for a CPG company, we used a multiple regression model because it allowed us to easily explain the variances in weather, seasonality, and holidays to our field team. Whether that makes sense for your business case is up to you.
From a pure academic perspective, generally in the past I’ve found ARIMA models to be the most accurate (KNIME has an ARIMA node), but more difficult to explain to the end user how it’s built (especially around ACF/PACF plots).
If you can provide more specific information about your goals of forecasting, the community can help you more.
This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.