Printing Feature Contributions in a Random Forest algorithm from the Treeinterpreter library leading to errors

Question

I am working on a dataset where I predict the risks of developing pancreatic cancer with respect to a number of variables. I have created a random forest, and want to find the feature contributions. I have already used the "Treeinterpreter" library, resulting in a contributions array that is three-dimensional. I want to display the contributions in the array beside the name of the factor/variable. I have used the code below to do so, however, the code responsible for displaying the contributions does not work. I have tried multiple methods, including converting the dataframe to a numpy array, and other methods such as .all() and .any(). However, none are producing the desired result.

What can be the right way to display the feature contributions with respect to each of the feature it represents?

     # -*- coding: utf-8 -*-
"""
Created on Mon Apr 15 13:39:19 2019

@author: GoodManMcGee
"""

import pandas as pd
from sklearn.metrics import accuracy_score
from sklearn import tree
from sklearn.model_selection import train_test_split
from sklearn import preprocessing
from sklearn.metrics import confusion_matrix
from sklearn.ensemble import RandomForestClassifier 
from IPython.display import Image
from sklearn.tree import export_graphviz
from treeinterpreter import treeinterpreter as ti
import matplotlib.pyplot as plt
import numpy as np
import itertools

data = pd.read_csv("pancreatic_cancer_smokers.csv")
target = data['case (1: case, 0: control)']
data.drop('case (1: case, 0: control)', axis=1, inplace=True)
x_train, x_test, y_train, y_test = train_test_split(data, target, test_size = 0.2)
clf = RandomForestClassifier(n_estimators=100)
clf.fit(x_train, y_train)
y_pred = clf.predict(x_test)
clf_accuracy = accuracy_score(y_test, y_pred)
clf_pred, clf_bias, contributions = ti.predict(clf, x_test)


#The code below was taken from DataDive's treeinterpreter tutorial. 
#The aforementioned messages applies to all code between the underscores
#///////////////////////////////////////////

for i in range(len(x_test)):
    print ("Instance", i)
    print ("Bias (trainset mean)", clf_bias[i])
    print ("Feature contributions:")
    for c, feature in sorted(zip(contributions[i], data.feature_names), 
                             key=lambda x: -abs(x[0])):
#An error occurs in the "data.feature_names" method in the code above:AttributeError: 'DataFrame' object has no attribute 'feature_names'. I have tried referenceing columns from datasets also, but that also leads to errors: ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
        print (feature, round(c, 2))
    print ("-"*20)   
#///////////////////////////////////////////

Answer 1

Try this in the last part of your code:

for i in range(len(x_test)):
    print ("Instance", i)
    print ("Bias (trainset mean)", clf_bias[i])
    print ("Feature contributions:")
    for c, feature in sorted(zip(contributions[i,:,0], data.columns),key=lambda x: -abs(x[0])):
        print (feature, round(c, 2))
    print ("-"*20)

The problem is that you are sorting contributions without taking into account that contributions is a 3D array and the column names is accesible with data.columns, not data.feature_names.