Getting started with Azure ml machine learning platform

Azure ml studio -getting started

Hi everybody, my name is shyambhu and I am going to provide you an introduction to Azureml and all its functionality. I will be posting details of working chances and specific details which experienced users generally do not provide. This post is normally targeted towards people who are not aware of Azureml or are just beginning to use it, but will also enrich people already working on it to find neat details and get more out of Azureml.
First of all, we need to know what is Azureml.

What is Azure Ml?

Azure Ml , i.e. Azure machine learning studio, is a platform created by Microsoft, to provide a all around platform for machine learning without much knowledge of the original codes used behind the machine learning algorithms. In Azureml, one can get free workspace from his/her microsoft account and then can access all of its functionality.

Signing into Azure ML:

First, if you have Microsoft account its good, otherwise first open a Microsoft account using one of your existing mail account and one phone no.

Secondly, please go to page

https://studio.azureml.net/ and on the right top corner there will be sign in option, which you have to then sign in through using your microsoft account and/or mobile no.

Homepage and basics:

Above is the picture of the first homepage. On the left there are the following options.

Experiments:

Experiments are environments where one can execute a lot of machine learning works. There are lot of tools available, which can be dragged, dropped and connected one to another, to create a work-flow or a full-phased machine learning model. More about that will come later.

Projects:

One or more than one experiments can be included in a project. There is a option in the bottom, named Add to project. One can select one or more experiments from the homepage of the idea-board and then add them to some existing project or a new project. This project option is important for gathering a number of experiments under one umbrella. Project can be used for long time works as well as working on a research oriented problem.

Web service:

There is the option called web service. One can create certain web services within some experiment, like prediction services and then deploy them and share them to third party via Azure Ml. These web services can be viewed and the dashboard and configuration as well as details as API keys and test versions can be seen about the web service from this option.

Notebooks:

Whenever one uploads a data set into Azureml, one gets a bunch of functions with it. One of them is opening it in a notebook, which is currently supported in python 3, python 2 and R. The notebook are Jupyter notebook. If someone does not know about jupyter, please refer to https://jupyter.org/, which is the official documentation of the jupyter notebook. Also, one can directly go into notebook, and open a notebook without associated with any dataset. Hence one can actually work on notebooks and machine learning projects without any software downloaded on the machine.

Datasets:

One can upload dataset from local computer, internet or azure-data storage houses. These uploaded datasets get copied in the cloud-space associated for the client, and then one can see all the datasets associated to a workspace under this dataset tag. Under this dataset, there are My dataset and samples options. Azure platform comes with a lot of sample data also. One can take these sample datas and explore their features and run experiments on them.

Trained models:

The last option on the right is trained models. All the trained models are listed under this option. One can explore a workspace and the available trained models.

On the top right corner there are different options available. Under the peoples' icon one can access forums. Under the smiling face sign, one can submit feedbacks. Under the questions sign one can take two actions i.e. taking a tour of Azureml platform, which includes opening a experiment, creating a model, scoring it and then at the end deploying it as a web service. Also one can visit the documentation of azureml which is created and maintained by Microsoft every year.

+New:

This is a option to import new (1)dataset (2) notebook (3) module (4) project and (5)Experiments. These options (4) and (5) will be rather delved into details in next blogs, as modules are interesting in themselves and can lead to a lot of technical and theoretical interesting topics.

This was the intro to Azureml. In the next blog we'll explore more of experiments, the options available and the corresponding theories if necessary.

If you are more of a book person than a blog one, still follow my blog for new and exciting contents, while you can read the book below:

20 Must-Know Math Puzzles for Data Science Interviews: Test Your Problem-Solving Skills

Introduction: When preparing for a data science interview, brushing up on your coding and statistical knowledge is crucial—but math puzzles also play a significant role. Many interviewers use puzzles to assess how candidates approach complex problems, test their logical reasoning, and gauge their problem-solving efficiency. These puzzles are often designed to test not only your knowledge of math but also your ability to think critically and creatively. Here, we've compiled 20 challenging yet exciting math puzzles to help you prepare for data science interviews. We’ll walk you through each puzzle, followed by an explanation of the solution. 1. The Missing Dollar Puzzle Puzzle: Three friends check into a hotel room that costs $30. They each contribute $10. Later, the hotel realizes there was an error and the room actually costs $25. The hotel gives $5 back to the bellboy to return to the friends, but the bellboy, being dishonest, pockets $2 and gives $1 back to each friend. No...

Machine learning and statistics with python

Search This Blog