Invalid Login Attempt

Lab: Forecasting Sales with Linear Regression in Azure Machine Learning Studio

Overview

In this lab, you will use AzureML to set up a model to forecast prices. We will also use R and RStudio in order to learn more R programming. Predicting the increase in sales from a number of factors is an example of regression, or you could simply call it scoring, which is a more familiar term.

If you want to know how small variations in input variables affect outcome, then you likely want to use a regression method. If you’re trying to predict scores, regression is likely a good choice for this business requirement.

There are different types of regression, and the selection of regression method depends on the business problem that you are trying to solve. For example, if you want to work out the probability that an object is in a given class, then you could use logistic regression, which is aimed at estimating class probabilities. In practical terms, what does that actually mean? Well, an example might be estimating the probability of fraud in a credit card purchase, where we might want to work out the probability that it is a fraudulent purchase.

We are also going to use a new method to work with missing data. In the Missing Data task, the PCA option approximates the covariance for the full dataset to reconstruct the missing data. In practice, this means that AzureML will use the PCA method to ‘guess’ what the missing data will be. For each column, AzureML will add an additional column which will identify whether the data was originally missing, or whether it was present. Later on, this makes it easier to visualize the data since we can include or exclude data which was originally missing, in line with the user requirements or to promote further analysis.


Details
  • Estimated time required to complete: 0 hours, 50 minutes
  • You will have access to this environment for 2 hours, 0 minutes
  • Learning Credits Required: 5
Who this lab is designed for
  • Data Professionals
  • Data Scientists
  • Data Analysts

Exercises

Exercise 1: Create a Machine Learning Studio Workspace

An Azure Machine Learning Studio Workspace allows you to use Machine Learning Studio to create and manage machine learning experiments and predictive web services. You can create multiple Workspaces, each one containing a set of your experiments, datasets, trained predictive models, web services, and notebooks. As the owner of a Workspace, you can invite other users to share the Workspace so you can collaborate with them on predictive analytics solutions.

In this exercise, you will create an Azure Machine Learning Studio Workspace.

Exercise 2: Upload the Dataset

Azure Machine Learning Studio is a powerful browser based visual drag-and-drop code free authoring environment for machine learning in Azure. It allows you to build, deploy and share predictive analytics solutions in a fully managed cloud service with minimal overhead and fast time to insights.

In this exercise, you will upload your dataset to Azure Machine Learning Studio.

Exercise 3: Forecasting prices with Azure Machine Learning Studio
In this exercise, you will use AzureML Studio clean and manage a dataset that refers to housing data of house prices in Boston. Once we have cleaned the data, we are going to use it in order to understand the predictors for prices. We will use the Linear Regression task, which can solve multiple inputs to predict a single numeric outcome. This is also known as multivariate linear regression.

Login to Start Lab


Not Registered? Already Registered?
Benefits
Real Time Labs allow you to learn technology in an isolated environment without the hassle or cost of setting up a dedicated learning environment.

How it works