Since I started this blog, I’m always wondering how I can improve it, what should I write about and how to apply ‘data science’ on anything F1-related. Today, I’m happy to announce that the qualifying and race predictions plus some additional insights (see further down for details) will be exposed via an API and they will be available in JSON format.
Before the summer break, F1 heads to Budapest for the Hungarian GP. It’s a very twisty track where passing another car is notoriously difficult. The top-4 teams are expected to occupy the first four rows with just one team per row, namely Mercedes, Ferrari, Red Bull and Force India. Bottas is predicted to start ahead of Hamilton although the latter has 5 pole positions in this circuit; the most of any current driver. It’s remarkable how little the order changes…
As I was brainstorming possible features for the ML model, I created a lot of graphs that helped me get a better understanding of what is hidden in the data and what features may be helpful for the ML algorithms. Here, I share some of the most interesting and weird ones!
After a race to forget for Lewis, F1 returns to Silverstone. Hamilton triumphed there last year in tricky conditions and I expect him to do the same this year; at least in qualifying. Bottas is expected to close a Mercedes 1-2 while the usual suspects (Vettel, Ricciardo, Kimi and Max) will follow in the next positions. It’ll be interesting to see if Stroll’s momentum carries on. Race predictions have been updated after the qualifying.
Following the most unpredictable Grand Prix so far this season, F1 is heading to the Red Bull Ring in Austria. The Mercedes are expected to occupy the first row off the starting line. Will this remain till the end of the race? I doubt so. Race predictions have been updated after the qualifying.
The new modelling approach used in the Canadian GP offered accurate predictions especially for the qualifying session. Following Mercedes’ 1-2 in Canada, a Ferrari is again expected to split the Mercedes in qualifying with Hamilton having 72.4% chances of starting in front of Vettel. However, Sebastian has a slightly higher probability (56.7%) to actually win the race. Race predictions have been updated after the qualifying.
I’m proud to announce that the predictions for the next Grand Prix are based on an entirely new modelling approach that promises significantly improved prediction accuracy. The new approach also offers deeper insight by giving for each pair of drivers the probability of the former starting or finishing above the latter. For the upcoming qualifying session, Hamilton edges Vettel by the slightest of margins (50.3% probability of Hamilton starting in front of Vettel). Race predictions have been updated after the…
Having explored the data, checked what is available and what’s not, found any inconsistencies and potential problems, it is time to get creative. Feature extraction is the step where an experienced data scientist can really make the difference and improve the subsequent model’s accuracy; many times more than it will be possible through algorithm selection and fine-tuning.
Monaco is going to be a very tight race with Mercedes, Ferrari and (why not) Red Bull since this twisty circuit will hide the deficiencies of their Renault engine. Hamilton and Vettel are again expected to fight for the pole position while Ricciardo may qualify above the Ferrari of Kimi. Jenson Button will struggle although Monaco may be an opportunity for McLaren to get some points. Race predictions have been updated after the qualifying.
With the Spanish Grand Prix approaching, here I present the forecast for the qualifying and race results. For some not-so-clear reason, the model seems to favour Daniel Ricciardo to start 2nd with Hamilton starting as low as 4th. This could due to his crash last year with Nico Rosberg; the model might have thought that it was Hamilton’s fault after all! Surprisingly enough, Verstappen is expected to start only 6th despite his impressive win in 2016.