There are many different issues with capturing seasonality in time series. In this short post, I’d like to discuss one of the most annoying ones. I’m talking about the seasonal pattern that shifts over time. What I mean is that, for example, instead of having the standard number of observations in the cycle (e.g., 24 […]
In forecasting, the term seasonality doesn’t always mean what you think it does. It encompasses more than just patterns repeating from one season to the next. In fact, seasonality covers a wide range of periodic behaviors, and can have some issues associated with the naming conventions. Should we discuss? First things first: when we say […]
If someone tells you that method X solves all problems and is the best one ever, they are either lying intentionally or do not fully understand what they are talking about. There is no such thing as “the best approach for everything”. Let me explain. Consider two products sold by retailers: ice cream and bread. […]
Image above depicts a fashion model making a presentation about a forecasting method. I like the forecast for the final period in that image… Over the last few years, I’ve seen phrases like “LightGBM model” or “Neural Network model” on LinkedIn many times, and the statistician in me shivers every time. So, I figured it’s […]
I’m currently doing a literature review for one of my papers on intermittent demand forecasting with machine learning, and I’ve noticed a recurring fundamental mistake in several recently published papers, even in respectable peer-reviewed journals. The mistake? Using error measures based on the Mean Absolute Error (MAE). This is a crime against the humanity when […]
One thing that bothers me when I read posts on social media or papers in peer-reviewed journals is the claim that a proposed approach is “assumption-free.” In forecasting, this is never true. Such an approach is like a spherical unicorn in a vacuum (see image above). Here’s why. Every model is a simplification of reality, […]
Christmas and the New Year are upon us, and I wanted to publish a celebratory post before taking a break. Instead of writing something educational, I decided to simply recommend a paper for you to read over the holidays – something you might have overlooked in the past couple of years. Here it is or […]
I’ve seen several times ML experts applying principles of classification for intermittent demand forecasting. For example, they try predicting, WHEN the demand will happen. This is not a very sensible thing to do. The featured image in this post shows two forecasting approaches: one that tries to predict when demand happens (the yellow line), and […]
While Naive is considered a standard benchmark in forecasting, there is a case where it might not be a good one: intermittent demand. And here is why I think so. Naive is a forecasting method that uses the last available observation as a forecast for the next ones. It does not have any parameters to […]
Anna Sroginis and I have been working on a new approach for intermittent demand classification over the past year. We’ve taken a fresh look at the problem, starting by asking: why do zeroes happen? Let’s discuss why indeed. First, a quick note: it’s a mistake to define intermittent demand simply as “demand with zeroes”. That […]