Book Image

Microsoft Azure Machine Learning

By : Sumit Mund, Christina Storm
Book Image

Microsoft Azure Machine Learning

By: Sumit Mund, Christina Storm

Overview of this book

Related Content you might be interested in

Current Title:

Small book image
Microsoft Azure Machine Learning
Sumit Mund | Christina Storm
Jun, 2015 | 212 pages
No titles found
Table of Contents (21 chapters)
Microsoft Azure Machine Learning
Microsoft Azure Machine Learning
Credits
Credits
About the Author
About the Author
Acknowledgments
Acknowledgments
About the Reviewers
About the Reviewers
www.PacktPub.com
www.PacktPub.com
Preface
Preface
Free Chapter
1
Introduction
Introduction
Introduction to predictive analytics
Machine learning
Introduction to Azure Machine Learning
Summary
2
ML Studio Inside Out
ML Studio Inside Out
Introduction to ML Studio
Getting started with Microsoft Azure
Inside ML Studio
Workspace as a collaborative environment
Summary
3
Data Exploration and Visualization
Data Exploration and Visualization
The basic concepts
Data exploration in ML Studio
Do it yourself
Summary
4
Getting Data in and out of ML Studio
Getting Data in and out of ML Studio
Getting data in ML Studio
Data format conversion
Getting data from ML Studio
Summary
5
Data Preparation
Data Preparation
Data manipulation
Splitting data
Do it yourself
The Apply SQL Transformation module
Advanced data preprocessing
Summary
6
Regression Models
Regression Models
Understanding regression algorithms
Train, score, and evaluate
Linear regression
Optimizing parameters for a learner – the sweep parameters module
The train neural network regression – do it yourself
Comparing models with the evaluate model
Other regression algorithms
No free lunch
Summary
7
Classification Models
Classification Models
Understanding classification
Training, scoring, and evaluating modules
Classifying diabetes or not
Predicting adult income with decision-tree-based models
Do it yourself – comparing models to choose the best
Multiclass classification
Multiclass classification with the Iris dataset
Multiclass classification with the Wine dataset
Summary
8
Clustering
Clustering
Understanding the K-means clustering algorithm
Creating a K-means clustering model using ML Studio
Do it yourself
Clustering versus classification
Summary
9
A Recommender System
A Recommender System
The Matchbox recommender
Building a recommendation system
Summary
10
Extensibility with R and Python
Extensibility with R and Python
Introduction to R
Introduction to Python
Why should you extend through R/Python code?
Extending experiments using the Python language
Extending experiments using the R language
Summary
11
Publishing a Model as a Web Service
Publishing a Model as a Web Service
Preparing an experiment to be published
Creating a scoring experiment
Specifying the input and output of the web service
Publishing a model as a web service
Consuming a published web service
Web service configuration
Updating the web service
Summary
12
Case Study Exercise I
Case Study Exercise I
Problem definition and scope
The dataset
Data exploration and preparation
Model development
Model deployment
Summary
13
Case Study Exercise II
Case Study Exercise II
Problem definition and scope
The dataset
Data exploration and preparation
Model development
Model deployment
Summary
Index
Index

Comparing models with the evaluate model

With the Evaluate Model, you can also compare two models side by side in the same experiment. The two input ports of the module can take the output from the two Score modules and generate the evaluation matrix to compare the output from the two inputs that the module accepts. As shown in the following screenshot:

Comparing models – the neural network and boosted decision tree

The Boosted Decision Tree Regression module is another regression algorithm that comes with ML Studio. It is an ensemble model like decision forest, but is a bit different.

Here, we will use both the Boosted Decision Tree Regression and Neural Network Regression modules in the same experiment and compare both using the Evaluate Model. We will use the Sweep Parameters module in both cases, to train the algorithm.

Create a new experiment and drag the same sample dataset—the Automobile price data (Raw) module. Then, use the two algorithms with the same training dataset (80 percent...

Previous Section
End of Section 7
Next Section