Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/apurva-modi/marchmadness
Kaggle's NCAA® ML Competition March Madness Competition presents a challenge to forecast the outcomes of all possible matchups in the Men’s and Women's Basketball Championships.
https://github.com/apurva-modi/marchmadness
challenge feature-engineering feature-extraction marchmadness ncaa-basketball python3 tensorflow
Last synced: 26 days ago
JSON representation
Kaggle's NCAA® ML Competition March Madness Competition presents a challenge to forecast the outcomes of all possible matchups in the Men’s and Women's Basketball Championships.
- Host: GitHub
- URL: https://github.com/apurva-modi/marchmadness
- Owner: apurva-modi
- Created: 2020-04-06T03:50:50.000Z (almost 5 years ago)
- Default Branch: master
- Last Pushed: 2020-04-10T14:26:21.000Z (almost 5 years ago)
- Last Synced: 2023-07-24T00:25:20.736Z (over 1 year ago)
- Topics: challenge, feature-engineering, feature-extraction, marchmadness, ncaa-basketball, python3, tensorflow
- Language: Jupyter Notebook
- Homepage:
- Size: 2.47 MB
- Stars: 0
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# marchmadness
Kaggle March Madness Competition - 2017 & 2019- This project presents a classical machine learning approach to predicting the NCAA Men’s Basketball Tournament. The project provides a good exposure to feature extraction and feature selection. We found that the most challenging part of the project is extracting and selecting the best features that will help in predicting the correct outcomes of the matches.
- In this work, we performed extensive feature extraction, feature selection and model selection to predict the outcome of march madness games . We confirmed that the simple logistic regression predicts the best outcomes with log loss of 0.48538.
- For the 2017 bracket, we correctly predicted 49/63 matches.
![alt text](2017bracket.jpg "2017PredictedBracket")
- For the 2019 bracket, we correctly predicted 46/63 matches.
![alt text](2019bracket.jpg "2019PredictedBracket")