You are on page 1of 9

ANALYSIS OF KOBE

BRYANTS
BASKETBALL
SHOTS
EEC 289Q SPRING 2016
ZIBRAN CHAUS

WHY?
Over half of NBA teams worth over 1 billion dollars
Large investment
Average Salary 4.3 million dollars
Kobes salary as of 2015 23 million dollars

Improved player training


Predictions for future game
Understanding your opponent
Win more

TOOLS
Language of choice Python
Xgboost
Gradient Boosting (Gradient Descent + Boosting)
Learning Technique for regression which makes prediction based on an ensemble of weaker prediction
models

Pandas
Data manipulation and analysis

Seaborn
Visualization library based on matplotlib

NumPy
Scientific computing, great for N-dimensional array objects

Kaggle
Testing Environment

METHODOLOGY
XGBoost has been used many times previously to win Kaggle
competition
One of the fastest gradient boosting algorithms
Can be multi-threaded

30698 total entries, 5000 blanks -> 25698 training data, 5000
testing data
25 Categories

METHODOLOGY
Picking which features matter

Gradient boosting parameters


Max Depth, N_estimators, Learning Rate

Cross-Validation and Fine tuning

DATA ANALYSIS
All Shots

Made Shots

Miss Shots

DATA ANALYSIS

CONCLUSION
Kaggle results based on return log loss
Kaggles Rank 1 Score : 0.57866
My Best score : 0.60262 (rank 83)

Machine Learning and Data Analysis in Python can be used to


improve a players strengths and find out what weakness they may
have.

WEBSITE

You might also like