选择 K 个最佳特征

Question

I was trying to find out the best features from a large dataframe. I am able to get the compressed dataframe values but I couldn't get the names of the features that were selected. Below is my code:

print('Shape of the bigramdf before feature selection:',bigram_df.shape)
if not os.path.isfile('smalldata/bigram_feather_top_100.feather'):
    SelectKBest(score_func=chi2,k=100).fit(bigram_df.iloc[:,:-1],df['class'])
    cols=SelectKBest.get_support(indices=False) # I am getting error here
    selc_k_best_byte_bigram=bigram_df[:,cols]
    selc_k_best_byte_bigram['id']=bigram_df['id']
    selc_k_best_byte_bigram.to_feather('smalldata/bigram_feather_top_100.feather')

    print('Shape of the bigramdf before feature selection:',selc_k_best_byte_bigram.shape)
else:
    selc_k_best_byte_bigram=pd.read_feather('smalldata/bigram_feather_top_100.feather')

I am getting the following error:

TypeError: get_support() missing 1 required positional argument: 'self'

Can some one help me in finding why I am getting this TypeError.

Answer 1

I think you need to initialise the class in a variable and then call .get_support. So try to replace:

SelectKBest(score_func=chi2,k=100).fit(bigram_df.iloc[:,:-1],df['class'])

with

k_best = SelectKBest(score_func=chi2,k=100).fit(bigram_df.iloc[:,:-1],df['class'])
cols = k_best.get_support(indices=False)