Microsoft Azure Machine Learning

What differentiates Microsoft Azure Machine Learning from other similar products?

I would spend so much time initially establishing my local environment with the correct versions of libraries but azure ml does all of that in the cloud. It is superior to the standard jupyter since it is attached to all of my data sources in the cloud and accessible compute power. It is the best because i can spin up a massive gpu instance with a single button press whenever i need to train a deep learning model.

What is your favorite aspect of this product?

I like how it manages to store all my history of experiments to make cross-comparing of the variation performance of different code versions simple. The terminal access is also convenient and useful as well when i need to install some custom packages not available in the default environment. The intellisense feature in the notebook editor is also quite surprisingly good and allow me to write faster with few mistakes. It is a well developed development environment.

What do you dislike most about this product?

It can be time consuming to compile the image of the environment. When your dependencies change on a regular basis you will be waiting a long time before the new image is created.

What recommendations would you give to someone considering this product?

This is a phenomenal upgrade to any person living in notebooks as it provides all the cloud power without installing it locally.

What differentiates Microsoft Azure Machine Learning from other similar products?

I do not (like many others do) love code writing all the little stuff and the designer that comes with this tool is precisely what i was in need of. It is superior to the rest in the way that you can easily connect and understand the module, i can see where the data is getting mixed up in the graphical representation.

What is your favorite aspect of this product?

The most satisfying aspect is the library of ready made modules to common operations such as principal component analysis or sentiment detectors. It just feels more akin to a visual orginator with the flexibility of deep customization. It has made my daily workflow more enjoyable and less frustrating.

What do you dislike most about this product?

Occasionally the error messages in the visual designer are a cube. One has to do some searching on what exactly module has failed in the pipeline.

What recommendations would you give to someone considering this product?

In case you need a visual approach to data science this is the most powerful lightweight on the market. It is not complex but simple to learn quickly.

What differentiates Microsoft Azure Machine Learning from other similar products?

other programs require you to write a hundred lines of code to test a couple of models but this software does that work on my behalf. it is the best because it automatically tests out different models and hyperparameters to find out which one is most accurate to my business problem. this saves me days of manual code and lets me meet my deadlines much sooner. it feels like a senior data scientist is right next to me giving me tips on which model to use on my business problem.

What is your favorite aspect of this product?

i like the way it manages the feature engineering aspect of the process without me necessarily having to do much with it. it also cleans up the missing data on its own which is a plus and the interface to view the results is also very user-friendly. the speed of training is also a huge boost to such a complicated task.

What do you dislike most about this product?

it is occasionally slow to supply the compute clusters with a little. i frequently have to wait five minutes to see the machines boot up.

What recommendations would you give to someone considering this product?

when you are in a hurry and require quality models quickly you must indeed have a go at this. it eliminates the trial and error in machine learning among busy people.