Azure Machine Learning For Sentiment Analysis

Azure ML Dashboard

 
In this article, I am going to explain how the Azure ML process can be used for sentiment analysis.
 
First thing we need to do is create a DataSet in My Datesets folder in Azure ML. I've uploaded a csv file in My Datasets folder and dragged that to the Dashboard.
 
Azure Machine Learning For Sentiment Analysis
 

Preprocessing of Data

 
The next step is the preprocessing of the text by selecting the “preprocess text”. In this step, we have selected the filed named as “review” as described in the in the assignment for preprocessing.
 
Azure Machine Learning For Sentiment Analysis
 
And connect the preprocess text with the csv file so that the csv file data will be used as input. We can select different option form the side pane so that the data can be processed as much as we need it.
 

Feature Hashing

 
The next step of preprocessed text is feature hashing. "Preprocess review” field has been generated after getting data from preprocessing  the text. That field will be used for feature hashing. I have selected hashing bit size as 10.
 
Azure Machine Learning For Sentiment Analysis
 

Filter Based Feature

 
The next step is to get the filter-based features selection. In the left side bar, expand Feature Selection and Filter Based Feature Selection. For that purpose I have selected the fish scoring method and the selected column is “sentiment” on which filter based feature will be implemented and selection will be done. I have implemented the filter-based feature selection method. The desired feature for my work is 500.
 
Azure Machine Learning For Sentiment Analysis
 

K Means Clustering

 
After a filter-based selection has been made, the next step is to apply clustering on the data and train the model. For training a model, two inputs are required, the input data and the clusterning. 
 
I've selected our previously preprocessed hashing data as data input and K means clustering.
 
Azure Machine Learning For Sentiment Analysis
 
Other properties are also set in the right pane. The single parameter is selected to train the model. The iteration value is set to 100. The input  is given from the K means clustering, and I set 20 clusters for my data.
 

Visualization

 
Now, we just need to train and run our model. We got the final visualization that looks like below. Based on our values selected earlier, you see there are 20 clusters represented by different colors. 
 
Azure Machine Learning For Sentiment Analysis
 
Here is the full visualization of the model.
 
Azure Machine Learning For Sentiment Analysis
 
In this article, we learned how to use Azure ML Studio to apply the K clustering method to generate a visual representation of data.