Previously we discussed how to measure accuracy of point forecasts and performance of prediction intervals in different cases. Now we look into the question how to tell the difference between competing forecasting approaches. Let’s imagine the situation, when we have four forecasting methods applied to 100 time series with accuracy measured in terms of RMSSE: […]
Theory of forecasting
Forecasting method vs forecasting model: what’s difference?
If you work in the field of statistics, analytics, data science or forecasting, then you probably have already noticed that some of the instruments that are used in your field are called “methods”, while the others are called “models”. The issue here is that the people, using these terms, usually know the distinction between them, […]
M-competitions, from M4 to M5: reservations and expectations
UPDATE: I have also written a short post on “The role of M competitions in forecasting“, which gives historical perspective and a brief overview of the main findings of the previous competitions. Some of you might have noticed that the guidelines for the M5 competition have finally been released. Those of you who have previously […]
What about all those zeroes? Measuring performance of models on intermittent demand
In one of the previous posts, we have discussed how to measure the accuracy of forecasting methods on the continuous data. All these MAE, RMSE, MASE, RMSSE, rMAE, rRMSE and other measures can give you an information about the mean or median performance of forecasting methods. We have also discussed how to measure the performance […]
How confident are you? Assessing the uncertainty in forecasting
Introduction Some people think that the main idea of forecasting is in predicting the future as accurately as possible. I have bad news for them. The main idea of forecasting is in decreasing the uncertainty. Think about it: any event that we want to predict has some systematic components \(\mu_t\), which could potentially be captured […]
Are you sure you’re precise? Measuring accuracy of point forecasts
Two years ago I have written a post “Naughty APEs and the quest for the holy grail“, where I have discussed why percentage-based error measures (such as MPE, MAPE, sMAPE) are not good for the task of forecasting performance evaluation. However, it seems to me that I did not explain the topic to the full […]
Comparing additive and multiplicative regressions using AIC in R
One of the basic things the students are taught in statistics classes is that the comparison of models using information criteria can only be done when the models have the same response variable. This means, for example, that when you have \(\log(y_t)\) and calculate AIC, then this value is not comparable with AIC from a […]
Naughty APEs and the quest for the holy grail
Today I want to tell you a story of naughty APEs and the quest for the holy grail in forecasting. The topic has already been known for a while in academia, but is widely ignored by practitioners. APE stands for Absolute Percentage Error and is one of the simplest error measures, which is supposed to […]
True model
In the modern statistical literature there is a notion of “true model”, by which people usually mean some abstract mathematical model, presumably lying in the core of observed process. Roughly saying, it is implied that data we have has been generated by some big guy with a white beard sitting in mathematical clouds using some […]