As stated in one of my previous posts, I will show the ratings that my model predicted for the 2021 Hungarian GP. This race was completely atypical in pretty much every sense, so unfortunately the performance of the model will be … questionable.
So I finally got the ratings for most of the media sources that I used to train the model. I couldn’t get the ratings for Autosport because apparently, you need to pay to see their ratings (what?). I have no problem with monetizing some aspects of a website. Sometimes there are great articles behind paywalls and I think they provide something extra to the user. Having said that, putting the ratings of the drivers behind a paywall? Seems unnecessary and frankly kind of dumb to me. Because of this, I will be removing Autosport from my predictive model for the next race, and instead, I will be substituting them with media outlets that publicly share their ratings. If you have any suggestions please let me know.
Regarding the model, the numbers are actually not that bad. I’m surprised to be 100% honest with you. Sure, we got very few predictions right, but the numbers, in general, seem to be quite close to the actual ratings.
Remember that the model right now doesn’t know anything about crashing into your rivals at turn 1, so it won’t be able to properly judge the performance of Valtteri Bottas or Lance Stroll. It also doesn’t know that Sebastian Vettel ended up in 20th place after being disqualified, so it also won’t be able to properly judge his performance.
The model wasn’t great for this race, but it wasn’t that bad either considering the circumstances. This race was as atypical as it gets.
Just as before, our model once again struggled the most with The Race predictions. They tend to be quite inconsistent and the predictions reflect that. The model predicted only one rating correctly—Alsonso’s 9—and only 40% of the predictions had an error of 1 point or less.
The model did a decent job at predicting the ratings of F1i and Planet F1. For both of these sources, the model had an error of 1 point or less in 70% of the cases which is quite decent.
One final thing to take into consideration when looking at the mean error is that it is quite skewed in this case. Most of the predictions for Bottas, Stroll and Vettel are quite off due to the strange circumstances of the race. Without those data points, our model would look quite better than it looks right now.
First of all, it’s important to mention that the table already considers the final results of the race, including Vettel’s demotion to 20th place. Penalties were not included in the original model, so unfortunately the computer has no way of handling these types of situations.
The model was trained on data from rounds 1 to 10, and unfortunately, it had never seen data as atypical as the one that it got from the Hungarian’s GP. Max Verstappen was finishing at the top of the grid pretty much all the time, so that’s all the model has seen. The predictions have some uncertainty, which I am not showing right here. Having said that, I can assure you that the uncertainty for these predictions is quite large.
On the good side, I will update this model after this race with the information from the 2021 Hungarian GP. This should make the model more robust, and hopefully, the quality of the predictions will improve.
In any case, I do not expect great results from this race but hey, sometimes that’s just how it goes. Whenever you see the results of a model on the internet, remember that people tend to post only the models that make “good” predictions, and hide the models that struggle. I’m not doing that here. This model will improve, and that’s part of the learning process.
First of all, thank you for taking the time to read this article. I will update this post once the ratings from the media outlets are released. I do not have high hopes for the predictions this time (ouch!) but I said that I was doing this post and I did that. Good or bad, I want to show you that the world of analytics is not perfect, and that sometimes there are obstacles on the road.
If you want to support me, there are multiple ways you can donate some money to help me keep this project alive. You can find all the options on either the about tab or the my supporters tab in the main menu.