Getting started with Azure Machine Learning and publishing the experiment as webservice

What is Azure Machine Learning?

Microsoft Azure Machine Learning is an easy Visual tool that allows you to make business ready applications in very less time .You can compare data ,visualize it and also easily use machine learning algorithms to train the dataset and get desired result. It helps you publish the experiment as Web service when you complete the workflow.

The  Azure ML Studio

The Azure ML Studio has lot of things to offer, demos that you can work upon and take reference from. Here in Azure ML Studio you can follow simulation examples and build on top of it. Using predictive analysis you can train a dataset and use prebuilt algorithm to create great adaptive solutions. As quoted Azure ML is predictive analytics with data driven Intelligence in the cloud.

The  Experiment scenario

The game of Poker

As per Wikipedia  "Poker is a family of gambling card games involving betting and individual play, whereby the winner is determined by the ranks and combinations of players' cards, some of which remain hidden until the end of the game. Poker games vary in the number of cards dealt, the number of shared or "community" cards, and the number of cards that remain hidden. The betting procedures vary among different poker games in such ways as betting limits and splitting the pot between a high hand and a low hand."

The condition

The intent of this challenge is automatic rules induction, i.e. to learn the rules using machine learning, without hand coding heuristics

Lets Start

Head to machine learning Azure ML

Then click on new 

http://4.bp.blogspot.com/-0CnXEGnvmIc/VWyB_rLrumI/AAAAAAAAAP4/kvMKt34L-qg/s320/Screenshot%2B%2528939%2529.png

                                                                                                                                            

Start with Machine learning then click on Quick Create

http://4.bp.blogspot.com/-h6Dd2owoOZA/VWyB_qhqS4I/AAAAAAAAAQM/Y38tL2z1wPE/s320/Screenshot%2B%2528940%2529.png

                                   Click on Blank Experiment

http://2.bp.blogspot.com/-y9S8geLF7bg/VWyB_r6NGuI/AAAAAAAAAQI/8r-j8cQ_ZG4/s320/Screenshot%2B%2528941%2529.png

Now you have to upload the data that will train for .I downloaded it from the Kaggle website  train.csv from your local machine.

http://1.bp.blogspot.com/-KtITMsoeibY/VWyCDQ0CqHI/AAAAAAAAAQc/Uc449UxlmeQ/s320/Screenshot%2B%2528942%2529.png

                                             As the data is uploaded you will get the green tick that data is uploaded

http://1.bp.blogspot.com/-5rlR9HJ52Wk/VWyCDRnEkfI/AAAAAAAAAQg/gfutND4IQWQ/s320/Screenshot%2B%2528943%2529.png

                                                                Now you need to project the columns which you need to work upon accordingly

http://4.bp.blogspot.com/-4zgbgnr03XU/VWyCDqY6kZI/AAAAAAAAAQk/sA7TF8ui8vE/s320/Screenshot%2B%2528944%2529.png

               In the next step you need to split the data then train the model using  Azure  ML algo I have used Multi Decision Forest, drag  and drop it to experiment area.

http://4.bp.blogspot.com/-OpMkUK5L6wg/VWyCI2Tqm7I/AAAAAAAAAQ4/LX8V8yRfbQA/s320/Screenshot%2B%2528945%2529.png

                                                                  Here you save the experiment and then visualize the data.

http://2.bp.blogspot.com/-NbVahj9JkBA/VWyCIeY1qzI/AAAAAAAAAQ0/EZwnmJGqCa0/s320/Screenshot%2B%2528946%2529.png

                           Here in now we need to Score the model so we key in score and drag and drop the score model  and connect the inputs.

http://4.bp.blogspot.com/-jSC6G8dfIZc/VWyCL8L5s1I/AAAAAAAAARE/4c-Iid7ZSEY/s320/Screenshot%2B%2528948%2529.png

                                               Upon this we need to evaluate the model.We will drag and drop the evaluate model.

http://2.bp.blogspot.com/-sS8V390gha0/VWyCNLu70QI/AAAAAAAAARU/1B4P0mza6pg/s320/Screenshot%2B%2528949%2529.png

                                         Now Run the experiment.Visualize the result right clicking the evaluate model and  see the result.

http://3.bp.blogspot.com/-jTdkULqdoVE/VWyCM97AuqI/AAAAAAAAARQ/vqZt0PaXB2c/s320/Screenshot%2B%2528950%2529.png

http://2.bp.blogspot.com/-fxhgEmMpmmc/VWyCOTmrUpI/AAAAAAAAARc/DYGqIt0dsBQ/s320/Screenshot%2B%2528951%2529.png

http://1.bp.blogspot.com/-f9Ie4PkQpgg/VWyCQrKfdzI/AAAAAAAAARo/hmERVIM-MdU/s320/Screenshot%2B%2528952%2529.png

                      Now it's time for us to create a Web service

                      Azure ML now will run the training data and  if everything is perfect will start scoring your experiment.The final result is a web service.

http://3.bp.blogspot.com/-obmGO9uVgVk/VWyCQ6jeTYI/AAAAAAAAARw/q-1XUrvhWzQ/s320/Screenshot%2B%2528953%2529.png

Azure ML will now run you through all the steps involved and the entire flow.A perfect GUI based flow will guide you what have been done.

http://3.bp.blogspot.com/-j-9qSHBOhg4/VWyCWU5_6oI/AAAAAAAAASI/E5-kh-hDEW0/s320/Screenshot%2B%2528960%2529.png

http://4.bp.blogspot.com/-jIVKTUiiFCg/VWyQD-SEccI/AAAAAAAAATM/gfZPAUZaMyU/s320/Screenshot%2B%2528958%2529.png

http://3.bp.blogspot.com/-QmnN2o1hl6o/VWyQC2bNE9I/AAAAAAAAATI/5F6siOz4W-Q/s320/Screenshot%2B%2528959%2529.png

                                             Now you can test the API of the experiment

http://1.bp.blogspot.com/-bYsEzZJdFao/VWyCWY2_AQI/AAAAAAAAASM/yE0kXejmWVw/s320/Screenshot%2B%2528961%2529.png

                                            If u click on test Azure ML will start testing your API

http://1.bp.blogspot.com/-AmJ7enHhXBI/VWyCZcSWNeI/AAAAAAAAASc/6DaVceBO40M/s320/Screenshot%2B%2528962%2529.png

http://1.bp.blogspot.com/-Vo3wY3W_MOI/VWyCZ9UmXYI/AAAAAAAAASg/Jk2QzdoleO8/s320/Screenshot%2B%2528963%2529.png

Lets recapitulate what we have done

1)Uploading the data(*.csv)file to Microsoft Azure

2)Then we projected the columns which we will work upon

3)Then we split the data in two halves(depending upon the relation we require)

4)In the next step we train the dataset with train model option(here in we select a column).

5)In the train model option we would like to implement an algorithm where in we implement Multicast decision forest algorithm.

6)We need to now score a model

7)After scoring ,the final step is evaluating the model.

8)Lastly we see how  easy it is implement to the entire model as a web service.

We see now how easy it is to use Azure Machine Learning to our needs for  predictive analysis of data.