preprocessing.py 文件源码

def fit(self, X, y=None):
        self._colmask = [True] * X.shape[1]
        self._colnames = X.columns.ravel().tolist()

        # Identify batches
        groups = X[[self.by]].values.ravel().tolist()
        self._colmask[X.columns.get_loc(self.by)] = False

        # Convert groups to IDs
        glist = list(set(groups))
        self._groups = np.array([glist.index(group)
                                 for group in groups])

        for gid, batch in enumerate(list(set(groups))):
            scaler = clone(self._base_scaler)
            mask = self._groups == gid
            if not np.any(mask):
                continue
            self._scalers[batch] = scaler.fit(
                X.ix[mask, self._colmask], y)

        return self