15.4 Creating Random Forests with Microsoft Azure Machine Learning Studio

The videos below show how to create Random Forests using Microsoft Azure Machine Learning Studio. If you would like to follow along, download the Mushrooms.csv data set and the CostSamsSales.csv data set and upload the data into Azure beforehand.

Random Forests for Classification

Microsoft Azure does a great job at making Random Forest models for classification. This is done using the Two-class Random Forest pill for outcome variables that only have two levels or the Multiclass Random Forest pill which is for outcomes that have multiple levels.

Random Forest for Regression

Microsoft Azure can create Random Forest models for regression, however, compared to other tools such as R and JMP, Azure's Regression Random Forest pill may not perform as well. Much of it depends on the data set.

In addition, Microsoft Azure has the Tune Model Hyperparameters pill which automatically selects the ideal parameters for many different models (E.g. Two-Class Random Forest for Classification). Tune Model Hyperparameters may not always be perfect, but it can be a good starting point.