Hi,
How to assess the prediction quality for future data. I know 95% CI , PI and in-sample testing.
I have my model, the AIC, BIC, etc. How a business manger can understand the MAPE/RMSE etc?
The question is if i made 7 days ahead predictions? How accurate my predictions? So that managers can arrange the required resources.
many thanks
T
It sounds like you have all the data you need, so you can be confident :)
Divide the data into two parts, a training set and a test set. Fit a model on the training set and get the forecasts. Check these forecasts values with the available data (test set) and determine the residuals. This is cross validation. See the blog of my PhD supervisor Rob J Hyndman (www.robjhyndman.com) for details about cross validation and forecast errors. Regards
The accuracy of your forecasting depends on the statistical model you used. There are many methods that one can use for forecasting so you have to decide which one is useful in your case. To see the accuracy of the model, diagnostic of the model (errors) can be helpful. The cross validation techniques (where you divide your available data in training and testing sets) can be used to assess the forecasting accuracy of the model. And finally, to tell to your manager about, e.g., MAPE, try to understand what it does. It basically gives the percentage average relative error (relative to the true value). let the actual value is 50 and the foretasted value is 48, this leads to the error of 2 (relative to 50) or simply 4% just for one day. Take the average of all days (in absolute form) and you will get your MAPE.
- Badges
- Science topic
- Similar topics
- Psychology