Classifier class
A class that trains a basic logistic regression or random forest model (as used in our testing) and returns the inputs required for the fair classification metrics method.
Note: Users do not have to use this class and can train their own classifiers.
| Function | Description |
|---|---|
logistic_regression | Trains a logistic regression model and returns real X data, real y data, and the predicted y data. |
random_forest | Trains a random forest model and returns real X data, real y data, and the predicted y data. |
Classifier
CLASS
Classifier()
init
Method
__init__(dataType='binary',inputDim=58,embeddingDim=32)
Initializes classifier with the following parameters
Parameters
- test_size [float]: test size for train_test_split
- random_seed [int]: random seed for consistent metrics
logistic_regression
Method
logistic_regression(df, df_real)
Fit a logistic regression model on df and predict on df_real. Split df_real into X data and y data.
Parameters
- df [DataFrame] - The synthetic/generated dataset using Fair Transformer GAN
- df_real [DataFrame] - The original dataset
Return type
- X_real [df]: Returns real x data from df_real. Can be fed into the fair classification method from the Metrics() class.
- y_real [np.array]: Returns real y data from df_real. Can be fed into the fair classification method from the Metrics() class.
- y_pred [np.array]: Returns the predicted y data from the logistic regression model predictions. Can be fed into the fair classification method from the Metrics() class.
random_forest
Method
random_forest(df, df_real)
Fit a random forest model on df and predict on df_real. Split df_real into X data and y data.
Parameters
- df [DataFrame] - The synthetic/generated dataset using Fair Transformer GAN
- df_real [DataFrame] - The original dataset
Return type
- X_real [df]: Returns real x data from df_real. Can be fed into the fair classification method from the Metrics() class.
- y_real [np.array]: Returns real y data from df_real. Can be fed into the fair classification method from the Metrics() class.
- y_pred [np.array]: Returns the predicted y data from the random forest model predictions. Can be fed into the fair classification method from the Metrics() class.