target_encoder_grid.py

class target_encoder(BaseEstimator, TransformerMixin):

    def __init__(self):
        pass

    def fit(self, X, y = None):
        return self

    def transform(self, X, y = None):
        #target encode lat and long
        #drop target
        qt = QuantileTransformer()
        X[['latitude','longitude']] = qt.fit_transform(X[['latitude','longitude']])
        X['latitude'] = pd.cut(X['latitude'], bins = [0,0.2,0.4,0.6,0.8,1], labels = [1,2,3,4,5])
        X['longitude'] = pd.cut(X['longitude'], bins = [0,0.2,0.4,0.6,0.8,1], labels = [1,2,3,4,5])
        X['location_gridnum'] = X['latitude'].astype(str) + str('_') + X['longitude'].astype(str)
        Xy = X.join(y)
        location_means = Xy.groupby('location_gridnum')['price'].mean()
        Xy['location_gridnum'] = Xy['latitude'].map(location_means)
        Xy['location_gridnum] = Xy[['latitude','longitude']].astype(float)
        X = Xy.drop('price', axis = 1)
        return X