Built Score Predictor using few Regression models to predict scores for each MLB team.

1. Set up a data science project structure in a new git repository in your GitHub account
2. Pick one of the game data sets depending your sports preference https://github.com/fivethirtyeight/nfl-elo-game https://github.com/fivethirtyeight/data/tree/master/mlb-elo https://github.com/fivethirtyeight/data/tree/master/nba-carmelo https://github.com/fivethirtyeight/data/tree/master/soccer-spi
3. Load the data set into panda data frames
4. Formulate one or two ideas on how feature engineering would help the data set to establish additional value using exploratory data analysis
5. Build one or more regression models to determine the scores for each team using the other columns as features
6. Document your process and results
7. Commit your notebook, source code, visualizations and other supporting files to the git repository in GitHub

• https://github.com/fivethirtyeight/data/tree/master/mlb-elo

• Use few Regression models to design a score predictor.

• We have transformed features, encoded categorical features, dropped unwanted features, treated missing values and removed incorrect data samples

• The coefficient R^2 is defined as (1 - u/v), where u is the residual sum of squares ((y_true - y_pred) ** 2).sum() and v is the total sum of squares ((y_true - y_true.mean()) ** 2).sum(). The best possible score is 1.0 and it can be negative (because the model can be arbitrarily worse). A constant model that always predicts the expected value of y, disregarding the input features, would get a R^2 score of 0.0.

• In statistics, the mean squared error (MSE) or mean squared deviation (MSD) of an estimator (of a procedure for estimating an unobserved quantity) measures the average of the squares of the errors—that is, the average squared difference between the estimated values and the actual value.

• In statistics, mean absolute error (MAE) is a measure of errors between paired observations expressing the same phenomenon. Examples of Y versus X include comparisons of predicted versus observed, subsequent time versus initial time, and one technique of measurement versus an alternative technique of measurement.

• We can say that Gradient Boost Regressor performed better than all others
• Although Neural Network(MLP) performed the worst, by tuning parameters we can make neural network to work better than all others. This is clear indication of over-fitting as we have left all parameters to default values.

• We could achieve quite good performance even without understanding lot about how these features actually worked.
• If we could understand how all the features affect score in-depth then we could make our models to predict much more accurately.
• we can also work a bit on Neural Network model by tuning parameters to achieve best results.

• Project description

• Contains link to download dataset.

• Jupyter Notebook for Exploratory data analysis, Visualization, Feature Engineering and Clustering.

• plot- Comparing Mean Squared Error for each model -mse_comparison.png
• Plot- Comparing Mean Absolute Error for each model -mae_comparison.png

• Info about Tools, frameworks and libraries required to reproduce the work flow