Case Study - Comparison between Two Class Boosted Decision Tree and Two Class Neural Network algorithm

Objective:

The objective of this study is to compare Two Class Boosted Decision Tree algorithm and Two Class Neural Network algorithm on the basis for certain data to predict the probability of a person to buy a bike or not on the basis of certain input parameters.

Solution:

Two different prediction models are build using “Two Class Boosted decision Tree Algorithm ” and “Two class Neural Network Algorithm”.

Two Class Boosted decision Tree is an ensemble learning method in which the second tree corrects for the errors of the first tree, the third tree corrects for the errors of the first and second trees, and so forth. Predictions are based on the entire ensemble of trees together that makes the prediction.

Two class Neural Network is a supervised learning method for classification, and therefore requires a tagged data set, which includes a label column. For example, you could use this neural network model to predict binary outcomes such as whether or not a patient has a certain disease, or whether a machine is likely to fail within a specified window of time.

Benefits:

1. Provide better understanding to determine which algorithm can provide best result in given situation.
2. Provides ability to take informed decision
3. Scalable model which can be modified to adjust more parameters if required
4. Both the models can be integrated to any existing system with little efforts.

Introduction

Two Class Boosted Decision Tree
Two-Class Boosted Decision Tree module in Azure Machine Learning Studio is used to create a machine learning model that is based on the boosted decision trees algorithm.
A boosted decision tree is an ensemble learning method in which the second tree corrects for the errors of the first tree, the third tree corrects for the errors of the first and second trees, and so forth. Predictions are based on the entire ensemble of trees together that makes the prediction.

Two-Class Neural Network
Two-Class Neural Network module in Azure Machine Learning Studio is used to create a neural network model that can be used to predict a target that has only two values.
Classification using neural networks is a supervised learning method, and therefore requires a tagged data set, which includes a label column. For example, you could use this neural network model to predict binary outcomes such as whether or not a patient has a certain disease, or whether a machine is likely to fail within a specified window of time.
After you define the model, train it by providing a tagged data set and the model as an input to Train Model or to Tune Model Hyper parameters. The trained model can then be used to predict values for new inputs.

Algorithm and Input Data:

Here we are using Microsoft Azure’s “Two Class Boosted Decision Tree” and “Two Class Neural Network” to predict the probability of a person to purchase a bike or not on the basis of certain input parameters.

Creating some data set, which contains all the information required to train a model, we are assuming that following input parameters may impact the prediction of possibility if the person will buy the bike or not.

Parameters:
1. MaritalStatusNumeric
2. GenderNumeric
3. Income
4. Children
5. EducationNumeric
6. CommuteDistanceNumeric
7. Cars
8. Age
9. HomeOwnerNumeric
10. Region
The value for the input parameters should be from the range or of type provided against the parameters listed below.

Parameters	Values
MaritalStatusNumeric	0 (Single)/1(Married)
GenderNumeric	0 (Male) / 1(Female)
Income	10000 -150000
Children	0-5
EducationNumeric	4 (Bachelors)
CommuteDistanceNumeric	1-5 (In Miles)
Cars	0-4
Age	25-89
HomeOwnerNumeric	0 (NO)/ 1(YES)
Region	Europe/ North America/ Pacific

Comparison

Here we are comparing “Two Class Boosted Decision Tree” and “Two Class Neural Network” by providing same data as input parameters and compared the output and the standard metrics by evaluating the scored model.
Sample Data

Input	Values
MaritalStatusNumeric	0
GenderNumeric	0
Income	10000
Children	0
EducationNumeric	4
CommuteDistanceNumeric	1
Cars	0
Age	25
HomeOwnerNumeric	1
Region	Europe

Output
For Two-Class Boosted Decision tree

For Two-Class Neural Network

Standard Metrics comparison
ROC and metric for Two-Class Boosted Decision tree

ROC and metric for Two-Class Neural Network

Conclusion:

By comparing “Two Class Boosted Decision Tree” and “Two Class Neural Network” and according to the standard metrics of evaluated model , we can a conclude that the accuracy for the “Two Class Neural Network” is more accurate than “Two Class Boosted Decision Tree”.