用于特征工程的分组和装箱数据

Question

I struggle to divide my data into bins for feature engineering. The data is Sale Price that I want to group by categorical data (Neighbourhood).

What am i doing wrong - I got NaN values for all the rows? Thanks!

    pricy_location = train['SalePrice'].groupby(train['Neighborhood']).mean()
    label = ['rank1', 'rank2', 'rank3', 'rank4', 'rank5']
    train['Pricy_Loc'] = pd.qcut(pricy_location, 5, labels=label, precision=2)
    train['Pricy_Loc'].head()

Answer 1

I think the problem arises because you are creating a dataframe grouped by neighborhood (which is only 25 rows long) and then trying to use the categories created for that dataframe and applying it to a much longer dataframe that is 1460 rows long. You can simply get the summarized data in a new column of your train dataframe and then bin the result:

train['Pricy_loc'] = train.groupby('Neighborhood')['SalePrice'].transform('mean')
label = ['rank1', 'rank2', 'rank3', 'rank4', 'rank5']
train['Price_loc_cat'] = pd.qcut(train['Pricy_loc'], 5, labels=label, precision=2)