Machine Learning -Part IV – Create a Machine Learning Model using Azure Machine Learning Studio


Sathish Nadarajan
SharePoint MVP
Published On :   05 Feb 2018
Visit Count
Today :  2    Total :   1336



In the earlier articles, we saw the introduction about the machine learning and various key concepts, terminologies of Machine learning. In this article, let us see, how to create a Demo Model on Azure Machine Learning Studio.

Before start creating the Demo, let us see What is Azure Machine Learning Studio is.

Azure Machine Learning Studio

· It is a Visual Editor for Composing, Testing, Refining and deploying Machine Learning models

· ML Studio Includes

Ø Hundreds of Modules

Ø Common Algorithms for Classification, Regression and more

· Supports numerous input formats

· The custom algorithms can be built any few languages like R and Python.

ML Studio simplifies machine learning by providing a drag-and-drop model in which you build workflow. With ML Studio and the rich of assortment of modules it offers for modeling workflow, you can often build sophisticated models without writing a single line of code. However, it allows you to insert R and Python code anywhere in the workflow, providing infinite flexibility in what you can model.

With this introduction, let us create a Demo. As usual, let us go by Step by Step procedure

1. Login to the Azure ML Studio Portal - https://studio.azureml.net/

clip_image002

2. Go to the Experiments Tab on the left quick launch and we could see the existing experiments.

3. Let us click on the New Button on the Bottom ribbon.

4. There are many prebuild templates are available. Let us create with a blank experiment.

clip_image004

5. Click on the Blank Experiment. We can see the blank Canvas as below.

clip_image006

6. Now, before proceeding, let us know, what we are going to do.

a. I am going to feed the model with the historical data of various cars.

b. Cleansing of the data to make it more meaningful

c. Split the data – 75% for the training and 25% for evaluation

d. Train the 75% of the splitted data with Linear Regression Algorithm

e. Evaluate the model

f. Deploy the model as web service.

g. Predict the Price of the Car.

7. With this, let us drag and drop the “Automobile price data” from the left hand side as show in the below image.

clip_image008

clip_image010

8. Search for Clean in the search experiment items and select the “Clean Missing Data”

clip_image012

9. Connect the Raw data to the Clean Data module. A simple mouse click and drag the link will do the below.

clip_image014

10. Now, Search for Select Columns in DataSet and drag and drop the item.

clip_image016

11. Connect the output of the cleaned data set to the Select Columns Item.

clip_image018

12. Now, select the “Select Columns in Dataset” and on the property pane, click launch column selector.

clip_image020

13. On the Popup, select the columns which we are going to consider for this model.

clip_image022

14. The final selected columns are as shown below.

clip_image024

15. Now, let us run the model to see the output as of now.

16. Click the Run button at the bottom Ribbon.

17. After the Run, we can visualize the dataset.

clip_image026

18. The Cleaned dataset will be as below.

clip_image028

We can export this dataset as well.

19. Add the “Split Data” function to the model and give the Fraction of rows as .75 on the right pane. i.e., 75% of the data should be splited to one leg and the rest 25 % on the other leg.

clip_image030

20. Then drag and drop the “Train Model”

clip_image032

clip_image034

21. On the Train Model Property Pane, select the Price column, because that is the value which we are going to predict.

clip_image036

22. Add the Score Model and Evaluate Model.

clip_image038

23. With this, we are done with creating the Model. Let us run the Model by clicking the Run on the Bottom.

24. The total time taken for training with the given data is less than a minute. This time will vary based on the volume of the dataset given as input.

clip_image040

In the upcoming article, let us see how to convert this as a web service and consume it from our Application.

Happy Coding,

Sathish Nadarajan.

Categories