python - why result of scaling each column always equal to zero? - Stack Overflow

I am using minmaxscaler trying to scaling each column. The scaled result for each column is always all zero. For example , below the values of df_test_1 after finishing scaling is all zero. But even with all values of zero, using inverse_transferm from this values of zero can still revert back to original values. But why the results of scaled are shown all zero?

from sklearn.preprocessing import MinMaxScaler
df_dict={'A':[-1,-0.5,0,1],'B':[2,6,10,18]}
df_test=pd.DataFrame(df_dict)
print('original scale data')
print(df_test)
scaler_model_list=[]
df_test_1=df_test.copy()
for col in df_test.columns:
   scaler = MinMaxScaler()
   scaler_model_list.append(scaler)  # need to save scalerfor each column since there are different if we want to use inverse_transform() later
   df_test_1.loc[:,col]=scaler.fit_transform(df_test_1.loc[:,col].values.reshape(1,-1))[0]

print('after finishing scaling')
print(df_test_1)
print('after inverse transformation')
print(scaler_model_list[0].inverse_transform(df_test_1.iloc[:,0].values.reshape(1,-1)))
print(scaler_model_list[1].inverse_transform(df_test_1.iloc[:,1].values.reshape(1,-1)))


original scale data
     A   B
0 -1.0   2
1 -0.5   6
2  0.0  10
3  1.0  18
after finishing scaling
     A  B
0  0.0  0
1  0.0  0
2  0.0  0
3  0.0  0
after inverse transformation
[[-1.  -0.5  0.   1. ]]
[[ 2.  6. 10. 18.]]

I am using minmaxscaler trying to scaling each column. The scaled result for each column is always all zero. For example , below the values of df_test_1 after finishing scaling is all zero. But even with all values of zero, using inverse_transferm from this values of zero can still revert back to original values. But why the results of scaled are shown all zero?

from sklearn.preprocessing import MinMaxScaler
df_dict={'A':[-1,-0.5,0,1],'B':[2,6,10,18]}
df_test=pd.DataFrame(df_dict)
print('original scale data')
print(df_test)
scaler_model_list=[]
df_test_1=df_test.copy()
for col in df_test.columns:
   scaler = MinMaxScaler()
   scaler_model_list.append(scaler)  # need to save scalerfor each column since there are different if we want to use inverse_transform() later
   df_test_1.loc[:,col]=scaler.fit_transform(df_test_1.loc[:,col].values.reshape(1,-1))[0]

print('after finishing scaling')
print(df_test_1)
print('after inverse transformation')
print(scaler_model_list[0].inverse_transform(df_test_1.iloc[:,0].values.reshape(1,-1)))
print(scaler_model_list[1].inverse_transform(df_test_1.iloc[:,1].values.reshape(1,-1)))


original scale data
     A   B
0 -1.0   2
1 -0.5   6
2  0.0  10
3  1.0  18
after finishing scaling
     A  B
0  0.0  0
1  0.0  0
2  0.0  0
3  0.0  0
after inverse transformation
[[-1.  -0.5  0.   1. ]]
[[ 2.  6. 10. 18.]]

• python
• pandas

Share Improve this question Follow asked Mar 3 at 3:41 roudanroudan 4,2601515 gold badges5050 silver badges9999 bronze badges

Add a comment  | 

1 Answer 1

Sorted by: Reset to default 1

According to MinMaxScaler DOC:

X : array-like of shape (n_samples, n_features) The data used to compute the per-feature minimum and maximum used for later scaling along the features axis.

When you reshape your data here:

df_test_1.loc[:,df_test.columns[1]].values.reshape(1,-1)

you get 1 row data with 4 columns in your case (and only 1 value in each of them) instead of 1 column with 4 rows .

You can fix your code one of the following ways:

1. Fix reshape dimensions: df_test_1.loc[:,col]=scaler.fit_transform(df_test_1.loc[:,col].values.reshape(-1,1))
2. Pass a dataframe instead of reshaped Series by indexing using list: df_test_1.loc[:,col]=scaler.fit_transform(df_test_1.loc[:,[col]])
3. Passing all required columns at once:

df_test_2=df_test.copy()
scaler = MinMaxScaler()
cols = df_test.columns
df_test_2.loc[:,cols]=scaler.fit_transform(df_test_2.loc[:,cols])