Can Machine Learning help F1 engineers predict an undercut attempt?🤔

@HearneLaurence contacted me about it, and later produced a model doing that!🤖

Solid lines: driver's pitting probability🔮
Dashed lines: the real pit.

The model predicted HAM pitting, and LEC covering it!💡

The plot has been taken from the Report of his Final Year Project.

The report is 114 pages long and very interesting: send him a message if you'd like to read it!

And if you're working for an F1 team you might be interested in his expertise.

Weekly reminder that you can JOIN the CHAT channel to discuss my analyses OUTSIDE of the 'comments'!
👇Use this link👇
https://t.me/FDataAnCHAT

Ferrari made TWO strategies work brilliantly: good strategy, tyre wear, and driving!👌

Check out the last stint (on ⚪️Hards): after pitting, SAI was 2s/lap quicker than LEC, who was managing his older tyres
LEC pushed more and more in that stint as the finish line got closer

Who will finish the season on top: Carlos Sainz or Charles Leclerc?🤔

Perez overtook both Mercedes cars in 2 laps, but overtaking Hamilton was way harder!

-HAM reached 303km/h;
-RUS just 292km/h (much weaker ERS deployment🪫).

Due to Russell's bad straight-line speed, PER could get the slipstream sooner, further increasing his top speed.

Remember the Machine Learning model that tried to predict undercut attempts?🤖

Fantastic news: all the scripts and the 114-page report have been uploaded to GitHub!😁

You will find it here👇
https://github.com/laurence9899/F1_Pitstop_Predict_ML

Kudos to @HearneLaurence for creating it!

Sainz's overtake was a real one - and Charles Leclerc didn't concede easily!😬

Sainz gained 0.5s over the straight, thanks to a better corner exit + slipstream and DRS.

Yet, he had to conclude the overtake into Turn 1: Leclerc braked several metres later than him!

(there was no team order to let Sainz by on the straight, only to not fight him hard)

Mercedes extended the first stint, losing ~2s/lap to the two McLarens for over 10 laps: mid-race they were just 8th and 9th.📉

Pace was excellent in the last stint (fresher tyres), but not enough to offset McL's lower tyre deg!🛞

Merc has multiple issues to solve. ⚠️

After pitting for a new set of Hards, Mercedes' pace suddenly improved by over 4s/lap, which is impressive even when taking into account the significant difference in tyre age.

McLaren's performance was less variable both between stints and throughout each stint as well.

Weekly reminder that you can find ALL my socials and extra content through THIS link
👇
https://linktr.ee/fdataanalysis

Weekly reminder that you can find ALL my socials and extra content through THIS link
👇
https://linktr.ee/fdataanalysis

🗒NOTES ON Chinese GP:

-First race in 5 years! First time with 18" rims🛞
-First sprint of 2024
-Similar to Bahrain, where Aston struggled and Merc edged McL. That was 4 races ago, though!
-2 stops are likely
-1.4km full-throttle section🚀

📸 @pirellisport

Your prediction? 🤔

The Chinese GP features a 1.4km full-throttle section!➡️Top speed: crucial for lap time and to attack/defend💡

Image: top speed achieved in the last qualifying, during each team's best lap.

🟢Aston has solved their drag issues, contrary to ⚫️Mercedes and 🟠McL.

🔴Ferrari will likely run the same rear wing used in all the races so far.

Chinese GP: Medium-high load wings

✅Downforce CRUCIAL from T1 to T13.

❌The 1.4km full-throttle section will punish excessive drag, especially if caused by the (fixed) lower plane!

🔵RBR maximizes DRS effectiveness through an unloaded lower plane + very loaded upper plane.

Significant differences among the teams: Mercedes. Ferrari, Alpine and Williams have a less loaded wing compared to Aston Martin, for example.

Compared to RBR, McL has a less loaded upper plane and more loaded lower plane, which should make their DRS less powerful.

MINISECTORS - FP1 ChineseGP
🟢STR was quickest on the main straight, and out of T3/T8/T12
🟠PIA fast in the twisty S1 and S3
🟡VER quickest in S3 (long straight)

Laptimes still very slow (almost 5s to last pole)➡️Hard to guess the running order!
Aston: 324km/h (+4km/h vs McL)

FASTEST SECTORS - FP1 ChineseGP

Mercedes/ALO on Hards
Others on Softs

🟠NOR quickest in the twisty S1 by far:
-McL is showing excellent grip
-Most teams are probably still far from the limits
He aborted the lap while in Sector 3

🟢STR: quickest in S2
🟡RBR: quickest in S3

Visualizing the effect of flexible wings!👀
(Collab with @justf1car 🤝)

Downforce and drag are proportional to the speed squared: they are very small at low speeds, and become huge at high speeds.

This is a problem: cars have less grip at lower speeds (when they need it the most) and face a lot of air resistance at high speeds (on the straights)!

The clever F1 engineers invented flexible wings to get a better compromise:
-As much downforce as possible at lower speeds➡️Cornering improves;
-Reduced drag at high-speed thanks to the flexion of the wing➡️Top speed improves!

Weekly reminder that you can JOIN the CHAT channel to discuss my analyses OUTSIDE of the 'comments'!
👇Use this link👇
https://t.me/FDataAnCHAT

TOP SPEEDS - SPRINT SHOOTOUT

Nine teams were in a narrow 3km/h range:
-Fastest (322km/h): Aston, Haas and Williams
-Slower: (319km/h): McL and Ferrari
-Slowest (just 315km/h): Alpine (high drag and, despite that, the drivers had to brake sooner than most!)

RACE PACE - Chinese GP SPRINT

🥇No one could match VER: 0.74s/lap quicker than anyone else!

🥈LEC was 2nd quickest, then🥉PER (+0.79s/lap) and ALO (+0.80s/lap). HAM was in the mix, too (+0.82s/lap)

McLaren was uncompetitive

What's your prediction for the race?🤔

Highest top speed in the Sprint: 348km/h for Stroll.

Leclerc was close (347km/h), and his average top speed was the highest together with Ricciardo's (337km/h for both).

Highest top speed without DRS: 324km/h (TSU).