tristantao/nba-predictor

Data Science Project for Predicting NBA Team Performance

NBA Predictor

Final project for CS194 (Intro to Data Science). Model for predicting NBA team performance.

Instructions For Scraping Data

Run the command with a link to a Yahoo NBA Box Score, such as http://sports.yahoo.com/nba/philadelphia-76ers-toronto-raptors-2014040928/

python yahoo_box_score_scraper.py <url>

Instructions for Setting Up Models

In analysis/ folder, set up the yhat config file as follows:

setwd('path to root dir of this repo')
source('analysis/yhatConfig.R')
yhat.config  <- c(
    username="your username",
    apikey="your api key here",
    env="http://sandbox.yhathq.com/"
)

To setup the Linear Model, in R, run lm.R by sourcing it.
Afterwards, go to deploy.R and run the segment on lm.

Similarly for SVM, source svm.R
To deploy run the relevant segment in deploy.R

Similarly for NaiveBayes, source naive_bayes.R
To deploy use the relevant segment in deploy.R

Python Wrapper (in candie_py_wrapper/)

The python wrapper is in the aforementioned folder.

You can call it via terminal:

python predict_game.py "New York" "Toronto" "2014-05-06" "lm"
#param: team1 team2 date model_used
#team1 is the expected home team.

You can also import predict_game.py and call the predict_game.run_model(...) to use it to better interate.