How to get text objects to work with sklearn classifier pipeline?

Question

Objective: to predict the probability of a given set of class with sklearn when model inputs are int, float, and objects (according to a pandas dataframe).

I am using the following dataset from UCI Repository: Auto Dataset

I have created a pipeline which almost works:

# create transformers for the different variable types.

from sklearn.pipeline import Pipeline
from sklearn.impute import SimpleImputer
from sklearn.preprocessing import StandardScaler, OneHotEncoder
import pandas as pd
import numpy as np

data = pd.read_csv(r"C:\Auto Dataset.csv")
target = 'aspiration'
X = data.drop([target], axis = 1)
y = data[target]

integer_transformer = Pipeline(steps = [
   ('imputer', SimpleImputer(strategy = 'most_frequent')),
   ('scaler', StandardScaler())])

continuous_transformer = Pipeline(steps = [
   ('imputer', SimpleImputer(strategy = 'most_frequent')),
   ('scaler', StandardScaler())])

categorical_transformer = Pipeline(steps = [
   ('imputer', SimpleImputer(strategy = 'most_frequent')),
   ('lab_enc', OneHotEncoder(handle_unknown='ignore'))])

# Use the ColumnTransformer to apply the transformations to the correct columns in the dataframe.
integer_features = X.select_dtypes(include=['int64'])
continuous_features = X.select_dtypes(include=['float64'])
categorical_features = X.select_dtypes(include=['object'])

import numpy as np

from sklearn.compose import ColumnTransformer

preprocessor = ColumnTransformer(
   transformers=[
       ('ints', integer_transformer, integer_features),
       ('cont', continuous_transformer, continuous_features),
       ('cat', categorical_transformer, categorical_features)])

# Create a pipeline that combines the preprocessor created above with a classifier.
from sklearn.neighbors import KNeighborsClassifier

base = Pipeline(steps=[('preprocessor', preprocessor),
                     ('classifier', KNeighborsClassifier())])

Of course, I would like to make use of predict_proba() which ends up giving me a bit of trouble. I tried the following:

model = base.fit(X,y )
preds = model.predict_proba(X)

However, I received an error:

ValueError: No valid specification of the columns. Only a scalar, list or slice of all integers or all strings, or boolean mask is allowed

Of course, here is the full traceback:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-37-a1a29a8b3623> in <module>()
----> 1 base_learner.fit(X)

D:\Anaconda3\lib\site-packages\sklearn\pipeline.py in fit(self, X, y, **fit_params)
    263             This estimator
    264         """
--> 265         Xt, fit_params = self._fit(X, y, **fit_params)
    266         if self._final_estimator is not None:
    267             self._final_estimator.fit(Xt, y, **fit_params)

D:\Anaconda3\lib\site-packages\sklearn\pipeline.py in _fit(self, X, y, **fit_params)
    228                 Xt, fitted_transformer = fit_transform_one_cached(
    229                     cloned_transformer, Xt, y, None,
--> 230                     **fit_params_steps[name])
    231                 # Replace the transformer of the step with the fitted
    232                 # transformer. This is necessary when loading the transformer

D:\Anaconda3\lib\site-packages\sklearn\externals\joblib\memory.py in __call__(self, *args, **kwargs)
    327 
    328     def __call__(self, *args, **kwargs):
--> 329         return self.func(*args, **kwargs)
    330 
    331     def call_and_shelve(self, *args, **kwargs):

D:\Anaconda3\lib\site-packages\sklearn\pipeline.py in _fit_transform_one(transformer, X, y, weight, **fit_params)
    612 def _fit_transform_one(transformer, X, y, weight, **fit_params):
    613     if hasattr(transformer, 'fit_transform'):
--> 614         res = transformer.fit_transform(X, y, **fit_params)
    615     else:
    616         res = transformer.fit(X, y, **fit_params).transform(X)

D:\Anaconda3\lib\site-packages\sklearn\compose\_column_transformer.py in fit_transform(self, X, y)
    445         self._validate_transformers()
    446         self._validate_column_callables(X)
--> 447         self._validate_remainder(X)
    448 
    449         result = self._fit_transform(X, y, _fit_transform_one)

D:\Anaconda3\lib\site-packages\sklearn\compose\_column_transformer.py in _validate_remainder(self, X)
    299         cols = []
    300         for columns in self._columns:
--> 301             cols.extend(_get_column_indices(X, columns))
    302         remaining_idx = sorted(list(set(range(n_columns)) - set(cols))) or None
    303 

D:\Anaconda3\lib\site-packages\sklearn\compose\_column_transformer.py in _get_column_indices(X, key)
    654         return list(np.arange(n_columns)[key])
    655     else:
--> 656         raise ValueError("No valid specification of the columns. Only a "
    657                          "scalar, list or slice of all integers or all "
    658                          "strings, or boolean mask is allowed")

Not sure what I am missing, but would appreciate any help possible.

EDIT: I am using sklearn version 0.20.

Answer 1

The error message points you in the right direction. The columns should be specified by name or index, but you pass the data columns as a DataFrame.

df.select_dtypes() doesn't output column indices. It outputs a subset of the DataFrame with the matched columns. Your code should be

# Use the ColumnTransformer to apply the transformations to the correct columns in the dataframe.
integer_features = list(X.columns[X.dtypes == 'int64'])
continuous_features = list(X.columns[X.dtypes == 'float64'])
categorical_features = list(X.columns[X.dtypes == 'object'])

so that the, for example, the integer columns get passed on as a list ['curb-weight', 'engine-size', 'city-mpg', 'highway-mpg']