Suppose I have the following classes, my questions are

1. is the size of tmp2 much larger than tmp1, as it needs to store extra df_x?
2. is there any strong reason of preferring Tmp2 over Tmp1? In this simple example, df_x had little processe, in other examples, I may want do a heavy process for df_x. I’m wondering how much extra burden would Tmp1 create .

@dataclass
class Tmp0:
    df_x: pd.DataFrame
    keep_zeros: bool = False

    def transform(self):
        if not self.keep_zeros:
            self.df_x = self.df_x.replace(0, np.nan).dropna(how='all').dropna(how='all', axis=1)
        return self.df_x


@dataclass
class Tmp1:
    df_x: pd.DataFrame
    df_y: pd.DataFrame
    keep_zeros: bool = False

    def __post_init__(self):
        self.df_x = Tmp0(self.df_x, self.keep_zeros).transform()

    def transform(self):
        self.df_y = self.df_y.where(self.df_x.notna())
        return self.df_y

@dataclass
class Tmp2:
    df_y: pd.DataFrame

    def __post_init__(self):
        self.df_x = Tmp0(self.df_x).transform()

    def transform(self, df_x, keep_zeros=False):
        df_x = Tmp0(df_x, keep_zeros).transform()
        self.df_y = self.df_y.where(df_x.notna())
        return self.df_y


df_x = load_data()
tmp1 = Tmp1(df_x, df_y)
tmp2 = Tmp2(df_y)