2

What is the best way to keep column names after doing one hot encoder in python? All my features are categorical so I did like below: so, after import the dataset it looks like below

 PlaceID       Date  ...  BlockedRet  OverallSeverity
0    23620  1/10/2019  ...           1                1
1    13352  1/10/2019  ...           1                1
2    13674  1/10/2019  ...           1                1
3    13501  1/10/2019  ...           1                1
4    13675  1/10/2019  ...           1                1

[5 rows x 28 columns]

after choosing the features, I want to transform them using one hot encoder because most of them are categorical, my question after doing that using: 

from sklearn.preprocessing import LabelEncoder, OneHotEncoder

hotencode = OneHotEncoder(categorical_features=[0])
features = hotencode.fit_transform(features).toarray()

enter image description here the result comes without original column names, how can I transform them with the same column name+0.,1,2,3.

user1941183
  • 21
  • 4
  • Sorry but to me it's very unclear what you are asking: 1) What does your DataFrame `df` looks like after loading the data? Please add an example in the question; 2) How is `df.iloc[:,+2:-1]` supposed to perform one-hot-encoding? To me it looks like just selecting the 3rd column; 3) What do you want to obtain at the end? Please share an example of your desired output in the question. – UJIN Nov 12 '19 at 10:22
  • hope the question now is clear, many thanks! – user1941183 Nov 12 '19 at 11:24

1 Answers1

6

Here is a simple example:

import pandas as pd

df = pd.DataFrame([
       ['green', 'Chevrolet', 2017],
       ['blue', 'BMW', 2015], 
       ['yellow', 'Lexus', 2018],
])
df.columns = ['color', 'make', 'year']

df

'''
    color       make  year  color_encoded  Color_0  Color_1  Color_2
0   green  Chevrolet  2017              1      0.0      1.0      0.0
1    blue        BMW  2015              0      1.0      0.0      0.0
2  yellow      Lexus  2018              2      0.0      0.0      1.0
'''

Approach 1: One Hot Encoder

from sklearn.preprocessing import LabelEncoder
le_color = LabelEncoder()
df['color_encoded'] = le_color.fit_transform(df.color)

from sklearn.preprocessing import OneHotEncoder
color_ohe = OneHotEncoder()

X = color_ohe.fit_transform(df.color_encoded.values.reshape(-1,1)).toarray()

dfOneHot = pd.DataFrame(X, columns = ["Color_"+str(int(i)) for i in range(X.shape[1])])
df = pd.concat([df, dfOneHot], axis=1)

df

'''
    color       make  year  color_encoded  Color_0  Color_1  Color_2
0   green  Chevrolet  2017              1      0.0      1.0      0.0
1    blue        BMW  2015              0      1.0      0.0      0.0
2  yellow      Lexus  2018              2      0.0      0.0      1.0
'''

Reference:

https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.OneHotEncoder.html

Approach 2: Get Dummies

df_final = pd.concat([df, pd.get_dummies(df["color"],prefix="color")], axis=1)


df_final

'''
    color       make  year  color_blue  color_green  color_yellow
0   green  Chevrolet  2017           0            1             0
1    blue        BMW  2015           1            0             0
2  yellow      Lexus  2018           0            0             1
'''

Reference:

https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.get_dummies.html

Suhas_Pote
  • 3,620
  • 1
  • 23
  • 38