Wednesday, 18 April 2018

Learn Machine Learning using Azure ML Studio - Part 1

   Azure Machine Learning (Azure ML) is a cloud service that helps people execute the machine learning process. As its name suggests, it runs on Microsoft Azure, a public cloud platform. Because of this, Azure ML can work with very large amounts of data and be accessed from anywhere in the world. Using it requires just a web browser and an internet connection.

In this sample, we will be using automobile data which will be used to predict the safety factor. Data consists of make, model and other details. Symboling will be our safety factor which we will be predicting in the sample with automobile data.
  1. You can access Machine learning studio @ https://studio.azureml.net/

2. Sign in to azure ML page

  1. Download the csv file from github  
  2. In the studio click on New button.

  1. Click on Dataset and upload the automobile csv file using From Local File and enter name of the dataset as automobiledata  
  2. Upload the saved csv file as below.


  1. After data is uploaded. Click New again and create Empty Project.

  1. Enter Name as below. Click on the Tick button to save.

  1. Click new blank Experiment by clicking New Button

  1. Rename experiment to “Automobile Data Regression” and search for created dataset(automobiledata) and then drag and drop to experiment canvas.  

  1. Search for "Select columns in dataset" in the  “Search experiment items” available in left toolbar.

  1. Drag and drop the module "Select columns in dataset" into the experiment canvas. Connect the output port of Automobiledata and “Select column in Dataset” column module as below.
  2. Select on the "Select columns in Dataset module" in experiment canvas and click on “Launch Column Selector”. In the popup, select the columns to be used for building the machine learning model. This process is also called as defining features(Inputs). We have selected all the columns in our data. We can also choose only particular columns that are needed.
  3. Search for “Linear Regression” module and drag into experiment canvas
  4. Search for “Split Data” module and drag into experiment canvas. Split data is used to split data for the training model and testing the model. In “Fraction of Rows ..” enter 0.75 which means we are training the model using the dataset with 75% of data and remaining 25% is used for testing the data. Connect "Split Data" as below.
  5. Search and Drag/Drop the "Train model"  into environment canvas and connect the “Linear Regression” algorithm module and "Split Data" module with “Train Model”.

  1. Select “Train Model” module and click on the “Launch Column”. We can now select the column that can be predicted. In our case it is “Symboling”(safety factor)

  1. Search and drop the score model into the experiment canvas. Connect the “Split Data” to “Score model” as below. Connect “train model” to Score model as below. A common use of scoring is to return the output as part of a predictive web service.
  2. Search and drop the "Evaulate model" into the experiment canvas as shown below.
  3. Save the model

To Run and deploy the model created navigate to Run & Deploy Azure ML


No comments:

Post a Comment

Build Bot using LUIS