Kaggle game done in MadHacks@Madison. What's the best strategy to win in PUBG? Should you sit in one spot and hide your way into victory, or do you need to be the top shot? This is what we are looking into!

What it does

It analyzes which independent variables have the most weight in determining finishing placement, and generates corresponding regression equation:
winPlacePerc = .498 * walkDistance - .239 * killPlace + .147 * weaponsAcquired + .096 * rideDistance + .114 * boosts
So, don't be too cautious! Walk around and find someone to fight:)

How we built it

We used a large data set which has about 18,000 pieces of data to train our regression model. We first used PCA to extract the main contributors of winning the game. Then we used these variables to run multi-variable regression. We Preprocessed data with Python and doing regression analysis with SPSS.

Challenges we ran into

Don't know how to select independent variables that are most relevant to the dependent variable in the process of regression analysis.

Accomplishments that we're proud of

Sketching out main independent invariables (strategies) in a number of them. Giving a practical implication for choosing future PUBG strategies that guarantee higher probability of winning:)

What we learned

Something about data analysis and something about machine learning.

What's next for Winner winner, chicken dinner!

Good question! It's just a kick-start of our fight (yes, fight) in Kaggle. Later we'll see whether we can predict Finish Placement accurately.

Built With

Share this project: