One-hot encoding credit data
It's time to prepare the non-numeric columns so they can be added to your LogisticRegression()
model.
Once the new columns have been created using one-hot encoding, you can concatenate them with the numeric columns to create a new data frame which will be used throughout the rest of the course for predicting probability of default.
Remember to only one-hot encode the non-numeric columns. Doing this to the numeric columns would create an incredibly wide data set!
The credit loan data, cr_loan_clean
, has already been loaded in the workspace.
This exercise is part of the course
Credit Risk Modeling in Python
Exercise instructions
- Create a data set for all the numeric columns called
cred_num
and one for the non-numeric columns calledcred_str
. - Use one-hot encoding on
cred_str
to create a new data set calledcred_str_onehot
. - Union
cred_num
with the new one-hot encoded data and store the results ascr_loan_prep
. - Print the columns of the new data set.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Create two data sets for numeric and non-numeric data
____ = ____.select_dtypes(exclude=['object'])
____ = ____.select_dtypes(include=['object'])
# One-hot encode the non-numeric columns
____ = pd.____(____)
# Union the one-hot encoded columns to the numeric ones
____ = pd.concat([____, ____], axis=1)
# Print the columns in the new data set
print(____.columns)