data_node_frame.py 文件源码

def make_unique_value_each_column (self, df, node_id):
        """ Dataframe? ??? ???? ??? ??? ?? ??? ???? 
            Unique Value return in Dataframe
        Args:
          params:
            * df : dataframe
            * node_id: nnid
        Returns:
            json
        Raises:
        """
        try:
            data_conf = dict()
            column_cate_unique = dict()
            numerics = ['int16', 'int32', 'int64', 'float16', 'float32', 'float64']
            for i, v in df.dtypes.iteritems():
                if (str(v) not in numerics):  # maybe need float
                    column_cate_unique[i] = df[i].unique().size
            data_conf['unique_cell_feature'] = column_cate_unique
            data_conf_json_str = json.dumps(data_conf)
            data_conf_json = json.loads(data_conf_json_str)
            return data_conf_json
        except Exception as e:
            logging.error("make_unique_value_each_column error : {0}, {1}".format(i,v))
            raise e