Maximizing expected return-MAD

Support
1

In MAD exercise python,weekly returns is used,I want to use mean_return how is the indexing?
tscv = TimeSeriesSplit(max_train_size=11, test_size=1)

Create a list to store mean returns with dates

for train_index, test_index in tscv.split(assets_):
train_assets = assets_.iloc[train_index]
test_assets = assets_.iloc[test_index]

    weekly_returns = train_assets.pct_change().fillna(method='bfill')
    mean_return = weekly_returns.mean()
    
    weekly_returns["Date"] = weekly_returns.index.format()
    weekly_returns.set_index("Date", inplace=True)