WORLD CUP ML MODEL

@sujar.tech

27.7K followers · usually 5.0K views a post

Views

100.0K

Likes

3.2K

Comments

55

Shares

174

SoundTikTok sound · original sound - sujar.tech

View velocity

+61 views over 20.3d · ~3/day · 6 observations

Promoted product

Football-Data APIaxonlearn.app

Axon teaches machine learning from scratch — from weights and gradients to transformers — built for developers who want to actually understand the math.

Why it worked

The video leverages the high interest in the World Cup and the intrigue of using machine learning for predictions. By breaking down specific match-ups and providing confidence percentages, it offers a unique and engaging perspective that appeals to both sports and tech enthusiasts.

Summary

The creator uses a machine learning model, specifically XGBoost, to predict World Cup match outcomes. They detail the data sources, features used, and provide predictions for several Round of 16 matches, including Canada vs. Morocco, France vs. Paraguay, Brazil vs. Norway, and England vs. Mexico. The creator also offers the code for their model on their app, Axon.

Structure

1Introduction to predicting World Cup winners with ML
2Overview of Round of 16 matchups
3Explanation of data sources and model type (XGBoost)

Original caption

We’ve gotten right all of our round of 32 predictions for the World Cup games so let’s see how the round of 16 turns out, we have Canada vs Morocco, Paraguay vs France, Brazil vs Norway and Mexico vs England I’m really hoping our World Cup machine learning model does well here and gets all the games right as well #coding #machinelearning #worldcup

More from @sujar.tech

Product placement

Football-Data API (acts: provides soccer data for professional teams, enabling users to fill gaps in their data. Removing it would mean the user has to manually find and compile this data, changing the workflow.)

Call to action

Download the app on the App Store or go to axonlearn.app

On-screen text

WORLD CUP ML MODEL machine the winners World Cup Round of 16 Tomorrow 1:00 PM Round of 16 Tomorrow 5:00 PM Round of 16 Tomorrow 9:00 PM Round of 16 Tomorrow 1:00 PM Round of 16 Tomorrow 5:00 PM Round of 16 Tomorrow 9:00 PM Round of 16 Sun, Jul 9 4:00 PM Round of 16 Sun, Jul 9 8:00 PM Round of 16 Sun, Jul 9 4:00 PM Round of 16 Sun, Jul 9 8:00 PM ETH 1 FT 0 MWI FEATURE_COLS = [ "home_for_avg", "home_xg_against_avg", "home_goals_for_avg", "home_shot_accuracy_pct", "home_possession_pct", "home_progressive_passes_pct", "home_confederation", "away_for_avg", "away_xg_against_avg", "away_shot_accuracy_pct", "away_possession_pct", "away_progressive_passes_pct", "away_confederation", "h2h_home_wins", "h2h_draws", "h2h_away_wins", "is_neutral" ] LaLiga CONMEBOL COPA AMERICA 2021 Axon - Learn Machine Learning Master Machine Learning Open Axon Learn ML the way it should be taught. Learn ML the way it should be taught. Download on the App Store GET IT ON Google Play axonlearn.app CAN 41% MOR 59% TO ADVANCE Match Prediction: CAN vs MOR (Advancement includes extra time + penalties) CAN 1.0 - 1.3 MOR EXPECTED GOALS EXPECTED SCORELINE: 0-1 1-0 PAR 23% FRA 77% TO ADVANCE Match Prediction: PAR vs FRA (Advancement includes extra time + penalties) FRA 0-2 1-1 1-0 0-1 12% 11% 10% 0% 32% 12% 9% 6% Match Prediction: BRA vs NOR (Advancement includes extra time + penalties) BRA 53% NOR 47% TO ADVANCE 1-1 1-0 0-1 12% 9% 6% Match Prediction: MEX vs ENG (Advancement includes extra time + penalties) MEX 45% ENG 55% TO ADVANCE 1-1 1-0 0-1 13% 8% 7%

Transcript

Let's use machine learning to predict the winners of the World Cup matches. So far, we've gotten right all of our round of 32 predictions, and this is going to be a really intense weekend. So let's predict the winners of Canada versus Morocco, Paraguay versus France, Brazil versus Norway, and Mexico versus England. To get the data for my model, I'm pulling it in primarily from the Football-Data API, and I'm filling in the gaps with the Football Data API. The type of model I built is an XGBoost model, which is basically a bunch of stacked decision trees, which help determine whether it'll be a win or a loss for this model. On 20 features across multiple tournaments, including La Liga, Copa America 2021, and many more. You guys have been asking for the code, so I put all of it on Axon. You can download the app on the App Store, and if you're on Android or desktop, go to axonlearn.app. For the Canada Morocco game, it's a really close one. The model is only 59% confident that Morocco will win with an expected score of 1-0. For the France Paraguay game, the model is super confident that France will win with a 77% probability and expected score of 2-0. For the Brazil Norway game, it's really close with Brazil just leading at 53% and the expected score is 1-1 of regular time and Brazil winning in either extra time or penalties. For one of the biggest games of the weekend, we have England versus Mexico, and England's just pushing ahead at 55% expected score of 1-1 at the end of regular time and they're in either extra time or penalties. Let me know what other games I want me to predict and follow.