New Knime user and new to the world of predictive modeling. Trying my first stab at building a forecasting model to forecast revenue for the next x amount of days.
I found this article https://www.knime.org/blog/all-you-need-is-the-lag-column-node and it was extremely helpful to build a prediction and compare my actual known data to the predicted data. I'm wondering now how do I take what this model has done and forecast or trend this out to predict what the next 30 days of revenue will be with no known data? Any help would be much appreciated.
You should be able to do this using the ARIMA Learner and Predictor nodes. If you don't have them already, you can install them from the ARIMA Extension available in KNIME Labs.
Thanks Roland. I found some similar posts that mentioneed using ARIMA as well that were helpful. I managed to add those to my workflow. I forecasted 31 days out, but when I've tried forecasting more than that I seem to get the same value for every time after that which seemed odd. I thought I would continue to see some variation of values based on the data. Am I missing something here?
if you do not have enough new input data, the series will just give the mean value as the best guess going forward.