Introduction to Azure Machine Learning

    Agum Junianteo

    This article will begin with a question, “Have you ever made a Machine Learning model?” If there is a nod or at least the “yes” answer appears in your mind, it means you could complement the knowledge that you have through this paper. But if you have never, then this article will give you new insights.

    Machine learning is a method of analyzing which aim to analyze, model, and make predictions based on data. This method is actually a branch of artificial intelligence, which is basically an attempt to understand how humans think and implement it through an algorithm.

    Nowadays, there are many programming languages that can be used to create Machine Learning models. However, many of them need a deep understanding or at least understand programming languages, such as the Python, R, or C ++. Sometimes it becomes an obstacle for people who come from scientific fields who do not study programming languages but understand Machine Learning algorithm.

    To overcome this issue, Microsoft Azure offers a solution in the form of Machine Learning Studio, which is a browser that uses a drag-and-drop feature that allows users to create a Machine Learning model where coding capabilities are not needed. This is possible with Azure Machine Learning (AzureML) in Machine Learning Studio.

    AzureML is a tool that can be used to carry out the process of making, testing, and presenting predictive analytics solutions. Machine Learning Studio will publish a model that has been created as web services and can be easily accessed by customer apps such as BI tools in Excel.

    In the fish story article, AzureML is located in a downstream position. If we relate to the fish story, then AzureML can be assumed as a chef who cooks and processes raw ingredients into a classy dish with good taste. However, AzureML is a dedicated tool only for processing data, so if you want to display the processed data, additional software is needed by utilizing web services features.

    Figure 1. AzureML interface

    On this occasion, an example of AzureML is used in predicting whether someone will buy an item from an ad based on several features, namely gender, age, and income. These following steps will show you how to use AzureML to create a model:

    1. First step
      Enter the dataset that we have, in the .csv format, into the system

      Figure 2. Dataset
      Figure 2. Dataset
    2. Second Step
      Check the data that has been entered

      Figure 3. Dataset Visualization
      Figure 3. Dataset Visualization 
    3. Third step
      Build a model using the drag-and-drop feature that has been provided by AzureML. In this example, the training model will be created using the “Two-Class Logistic Regression” algorithm. The location of the function is shown by the image below.

      Figure 4. Model
      Figure 4. Model
    4. Step Four
      Run the program by clicking the Run button
    5. Step Five
      Evaluate the model that has been made, whether the results are optimal or need to be turned back to the model parameters.

      Figure 5. Model Evaluation
      Figure 5. Model Evaluation
    6. Step SixIf a Web Services function is needed, it can be done using the Set Up Web Service feature which can later be linked to Microsoft Excel.
      Figure 6. Web Services Features
      Figure 6. Web Services Features
      Figure 7. Excel as Web Services
      Figure 7. Excel as Web Services

      Based on the example of using AzureML above, we can see that making a Machine Learning model can be done without the need for the ability of a particular programming language, simply by understanding the flow of data entry, modeling data, and testing data. Moreover, AzureML can also be connected with other applications such as Microsoft Excel

    Leave a comment

    Your email address will not be published.