Artificial Neural Networks with Net# in Azure ML Studio

The ideas for neural networks go back to the 1940s. The essential concept is that a network of artificial neurons built out of interconnected threshold switches can learn to recognize patterns in the same way that an animal brain and nervous system does.

Though the name “neural network” gives an idea of a ‘black box’ type predictive operation; ANN is a set of mathematical operations.

VqOpE

As the name implies by itself; neural network is a structural ‘network’. The nodes of the neural network are organized in layers and the nodes are connected with each other with edges. The edges are directional and they are weighted.

Azure Machine Learning Studio comes with pre-built neural network modules that can easily use for predictive analytics.

NN models

Pre-built neural networks in AML Studio  

Multiclass Neural Network Module –

Used for multiclass classification problems. The number of hidden nodes, the learning date, number of learning iterations and many parameters can be changed easily by changing the module properties.

Two-Class Neural Network –

Ideal for binary classification problems. Same as the Multiclass neural network module, the properties of the neural network can be changed by the module properties.

Neural Network regression –

This is a supervised machine learning method that can be used to predict a numerical value.

These simple pre-built modules can be added to your ML experiment with just a drag and drop and change the parameters by changing the module properties. What you going to do if you want to implement a complex neural network architecture? Or to create a deep neural network with more hidden layers?

AzureML Studio comes handy here with providing you the ability to define the hidden layer/layers of the ANN with a script. Net# scripting language provide the ability to define almost any neural network architecture in an easy to read format.

Net# scripting language is able to

  • Create hidden layers and control the number of nodes in each layer.
  • Specify how layers are to be connected to each other.
  • Define special connectivity structures, such as convolutions and weight sharing bundles.
  • Specify different activation functions.

In Azure Machine Learning, you can add the Net# scripts by choosing ‘Custom definition script’ in Hidden layer specification property. By default, it would set to the fully connected case.

properties

Net# lexical is more similar to C#. The structure of a Net# script has four main sections.

  1. Constant declaration (Optional) – Define values used elsewhere in the neural network definition
  2. Layer declaration – The input, hidden and output layers are defined with the layer dimensions. The layer declaration for hidden or output layer can include the output function.
  3. Connection declaration – You can define connection bundles (Full, Filtered, Convolutional, Pooling, Response normalization) – Full connection bundle is the default configuration.
  4. Share declaration (Optional) – Defining multiple bundles with shared weights.

This is a simple neural network defined by a Net# script to perform a binary classification. You can customize the number of hidden neurons and the activation functions and see how the accuracy of the model variate.

<!– HTML generated using hilite.me –>

//A simple neural network definition
//auto keyword allows the ANN to automatically include all feature columns in the input examples
//input layer named Data
input Data auto;

//Hidden layer named "H" including 200 nodes
hidden H [200] from Data all;

//output layer named "Out" including 2 nodes (binary classification problem) 
//Sigmoid activation function has been used.
output Out [2] sigmoid from H all;

For more insides here’s the resources – https://docs.microsoft.com/en-us/azure/machine-learning/studio/azure-ml-netsharp-reference-guide#overview

Advertisements

Evaluating AzureML Experiments

Azure Machine Learning Studio allows you to build and deploy predictive machine learning experiments easily with few drags and drops (technically 😉).

The performance of the machine learning models can be evaluated based on number of matrices that are commonly used in machine learning and statistics available through the studio. Evaluation of the supervised machine learning problems such as regression, binary classification and multi-class classification can be done in two ways.

  1. Train-test split evaluation
  2. Cross validation

Train-test evaluation –

In AzureML Studio you can perform train-test evaluation with a simple experiment setup. The ‘Score Model’ module make the predictions for a portion of the original dataset. Normally the dataset is divided into two parts and the majority is used for training while the rest used for testing the trained model.

train-test

Train-test split

You can use ‘Split Data’ module to split the data. Choose whether you want a randomized split or not. In most of the cases, randomized split works better. If the dataset is having a periodic distribution for an example a time series data, NEVER use randomized split. Use the regular split.

Stratified split allows you to split the dataset according to the values in the key column. This would make the testing set more unbiased.

  • Pros-
    • Easy to implement and interpret
    • Less time consuming in execution
  • Cons-
    • If the dataset is small, keeping a portion for testing would be decrease the accuracy of the predictive model.
    • If the split is not random, the output of the evaluation matrices are inaccurate.
    • Can cause over-fitted predictive models.

Cross Validation –

Overcome the mentioned pitfalls in train-test split evaluation, cross validation comes handy in evaluating machine learning methods. In cross validation, despite of using a portion of the dataset for generating evaluation matrices, the whole dataset is used to calculate the accuracy of the model.

K-fold_cross_validation_EN

k-fold cross validation

We split our data into k subsets, and train on k-1 of those subsets. What we do is holding the last subset for test. We’re able to do it for each of the subsets. This is called k-folds cross validation.

  • Pros –
    • More realistic evaluation matrices can be generated.
    • Reduce the risk of over-fitting models.
  • Cons –
    • May take more time in evaluation because more calculations to be done.

Cross-validation with a parameter sweep –

I would say using ‘Tune model Hyperparameters’ module is the easiest way to identify the best predictive model and then use ‘Cross validate Model’ to check its reliability.

Here in my sample experiment I’ve used the breast cancer dataset available in AzureML Studio that normally use for binary classification.

experimentThe dataset consists 683 rows. I used train-test split evaluation as well as cross validation to generate the evaluation matrices. Note that whole dataset has been used to train the model in cross validation case, while train-test split only use 70% of the dataset for training the predictive model.

Two-class neural networks has used as the binary classification algorithm. The parameters are swapped to get the optimal predictive model.

When observing the outputs, the cross-validation evaluation provides that model trained with whole dataset give a mean accuracy of 0.9736 while the train-test evaluation provides an accuracy of 0.985! So, is that mean training with less data has increased the accuracy? Hell no! The evaluation done with cross-validation provides more realistic matrices for the trained model by testing the model with maximum number of data points.

Take-away – Always try to use cross-validation for evaluating predictive models rather than going for a simple train-test split.

You can access the experiment in the Cortana Intelligence Gallery through this link –

https://gallery.cortanaintelligence.com/Experiment/Breast-Cancer-data-Cross-Validation

Copying & Migrating AzureML experiments

A set Major advantages in using cloud based machine learning platforms are the ability of collaborative projects, easy sharing and easy migration.  Within AzureML Studio you can share or migrate the experiments using various approaches.

01. Share AzureML workspace

If you want to share all the experiments in your workspace with another user, this is the best option you can go with. All your built experiments, trained models, datasets would be shared with the users with this permission.

  1. Click SETTINGS in the left pane
  2. Click the USERS tab
  3. Click INVITE MORE USERS at the bottom of the page

ml4The users you inviting should have a Microsoft account or a work/school account from Azure Active Directory. Two user access levels can be defined as “Users” and “Owners”.

02. Copy experiment to an AzureML workspace

If you want to migrate an experiment from the current workspace to another, you can go for the experiments pane and click “Copy to workspace”. Note that you only can copy experiments to the workspaces in the same Azure region. This is important if you want to move your experiment from a free tier workspace to a paid standard tier.

ml6You’ll not be able to copy multiple experiments using a single click. If you have such kind of scenario, use poweshell scripts as instructed in this descriptive post.

03. Publish to Gallery

ml7For me this is one of the most useful options. You can use this option in two ways. One is to make the experiments public and in a way that only accessible through a shared link. If you share the experiment publicly that will be listed in the Cortana Intelligence Gallery.

ml8If you want to share an experiment only with your peer group, publishing as an ‘unlisted’ experiment is the best way. Users can open the experiment in their own AzureML studio. This option can be used to migrate your experiment within different workspaces as well as between different azure regions. Only the users who’s having the link you shared can only view or use the experiment you shared.

Simple Linear Regression with Azure ML + Python

1419973816879Simple linear regression is a statistical method that allows us to summarize and study relationships between two continuous (quantitative) variables: One variable, denoted x, is regarded as the predictor, explanatory, or independent variable. The other variable, denoted y, is regarded as the response, outcome, or dependent variable.

Typically when we doing regression analysis, we consider about the correlation of coefficient of the input variables. Correlation analysis measures the extent to which two variables vary together, including the strength and direction of their relationship.

correlation_dot_graphsLinear correlation coefficient(also called Pearson product-moment correlation coefficient) measure of the strength and direction of a linear association between two random variables.

I used the Istanbul Stock Exchange dataset to demonstrate the steps in doing a simple linear regression prediction. Azure Machine Learning experiment has built (get the experiment from here) for building the regression model. Built-in Bayesian Linear Regression algorithm has been used for building the model.

capture1The most interesting part is coming with python! 🙂

I’ve used a Jupyter Notebook and fetched the data to that workspace to visualize the dataset and to calculate the coefficient values between each variable. Pearsonr method in scipy library has used for that.

Refer the iPython notebook from Azure Notebook for the complete python script and the visualizations.

https://notebooks.azure.com/library/Python%20Visualizations/html/Istanbul%20Stock%20Python%203%20notebook.ipynb

Do run the code by your own. You’ll get it for sure!

 

Jupyter Notebook on AzureML

plot_regression_3d_1 If you are fond of playing with data to dig out the relationships of it and to plot interesting visualizations with data; python is the language you should speak.

Over the years, with the strong community support, python language got dedicated libraries for data analysis and predictive modeling like scikit-learn, Tensorflow, Theano etc. Even the ultimate IDE in town; Visual Studio started supporting python! So, no hesitation. Python is a great choice to make.

You can use many IDEs or even a simple text editor to write your python files. But python comes with a handy web application; Jupyter notebook that can be used to do your code. Even compile it!

Jupyter gets its birth in 2014 as a spin-off project of IPython; which is a command shell for interactive computing in multiple programming languages, originally developed for the Python.

Why Jupyter?

Jupyter notebook is a very popular tool among data scientists which as a web application that allows you to create and share documents that contain live code, equations, visualizations and explanatory text. “Jupyter” is a loose acronym meaning Julia, Python and R. One of the most prominent uses you get when using Jupyter notebook is the ability of sharing the data transformation and visualization steps with your peers.

If you want to run Jupyter notebook in your local machine do refer the link below. With a few easy steps, you can have Jupyter notebook up and running in your machine.

http://jupyter.readthedocs.io/en/latest/install.html

One of the easiest ways to use Jupyter is running the notebook on Azure. No need to have python or the dependencies of it installed on your local machine. You can create, edit and share the Jupyter notes using Azure Machine Learning Studio. All the execution happens on the cloud.

Let’s get started!

1Access your notebook from “Notebooks” tab of AzureML Studio. When creating a new notebook, you can select which language and version you want to have in your notebook. Python 2, Python 3 and R are the supported languages right now.

Same as the Jupyter notebook running on the local machine, you get the same IPython interface on your browser.

2On the notebook menu bar, you can find out the ‘help’ menu which contains a brief user interface tour as well as a list of keyboard shortcuts that you can use to drive the notebook.

Here’s a little data mashup I’ve done using the famous ‘Iris dataset’ included in python sklearn. The .ipynb file is available on my github repo. Feel free to download and play with. A static html page created with the notebook output also included in the repo.

Azure is coming up with Azure Notebook preview feature. Here’s Iris visualization hosted on Azure Notebook

https://notebooks.azure.com/library/Python%20Visualizations/html/Iris+Data+Visualization.ipynb

No Machine learning algorithms or complex code snippets here. Just a data visualization & data transformation. 🙂

 

 

 

Time Series Forecasting with Azure ML

airline1_web-0When we have a series of data points indexed in time order we can define that as a “Time Series”. Most commonly, a time series is a sequence taken at successive equally spaced points in time. Monthly rainfall data, temperature data of a certain place are some examples for time series.

In the field of predictive analytics, there are many incidents that need to analyze time series data and forecast the future values of that based on the previous values. Think of a scenario where you’ve to do a time series prediction for your business data or an incident where part of your predictive experiment contains a time series field that need to predict the future data points… There are many algorithms and machine learning models that you can use for forecasting time series values.

Multi-layer perception, Bayesian neural networks, radial basis functions, generalized regression neural networks (also called kernel regression), K-nearest neighbor regression, CART regression trees, support vector regression, and Gaussian processes are some machine learning algorithms that can be used for time series forecasting.

See here for more about these methods

Autoregressive Moving Average (ARIMA), Seasonal-ARIMA, Exponential smoothing (ETS) are some algorithms that widely used for this kind of time series analysis. I’m not going to dig deep into the algorithms, trend analysis and all numbers & characteristics bound with time series. Just going to demonstrate a simple way that you can do time series analysis in your deployments using Azure ML Studio.

After adding a dataset that contains a time series data into AzureML Studio, you can perform the time series analysis and predictions by using python or R scripts. In addition to that ML Studio offers a pre-built module for Anomaly detection of time series datasets. It can learn the normal characteristics of the provided time series and detect deviations from the normal pattern.

Here I’ve used forecast R package to write code snippets enabling AzureML Studio to do TS forecasting using popular time series algorithms namely as ARIMA, Seasonal ARIMA and ETS.

ARIMA seasonal & ARIMA non-seasonal

#ARIMA Seasonal / ARIMA non-seasonal 
library(forecast)
# Map 1-based optional input ports to variables
dataset1 <- maml.mapInputPort(1) # class: data.frame
dataset2 <- maml.mapInputPort(2) # class: data.frame

#Enter the seasonality of the timeseries here
#For non-seasonal model use '1' as the seasonality
seasonality<-12
labels <- as.numeric(dataset1$data)
timeseries <- ts(labels,frequency=seasonality)
model <- auto.arima(timeseries)
numPeriodsToForecast <- ceiling(max(dataset2$date)) - ceiling(max(dataset1$date))
numPeriodsToForecast <- max(numPeriodsToForecast, 0)
forecastedData <- forecast(model, h=numPeriodsToForecast)
forecastedData <- as.numeric(forecastedData$mean)

output <- data.frame(date=dataset2$date,forecast=forecastedData)
data.set <- output

# Select data.frame to be sent to the output Dataset port
maml.mapOutputPort("data.set");

 

ETS seasonal & ETS non-seasonal

#ETS seasonal / ETS non-seasonal 
library(forecast)
# Map 1-based optional input ports to variables
dataset1 <- maml.mapInputPort(1) # class: data.frame
dataset2 <- maml.mapInputPort(2) # class: data.frame

#Add the seasonality here
#Assign seasonality as 'a' for non-seasonal ETS  
seasonality<-12
labels <- as.numeric(dataset1$data)
timeseries <- ts(labels,frequency=seasonality)
model <- ets(timeseries)
numPeriodsToForecast <- ceiling(max(dataset2$date)) - ceiling(max(dataset1$date))
numPeriodsToForecast <- max(numPeriodsToForecast, 0)
forecastedData <- forecast(model, h=numPeriodsToForecast)
forecastedData <- as.numeric(forecastedData$mean)

output <- data.frame(date=dataset2$date,forecast=forecastedData)
data.set <- output

# Select data.frame to be sent to the output Dataset port
maml.mapOutputPort("data.set");

 

The advantage of using R script for the prediction is the ability of customizing the script as you want. But if you want looking for an instant solution for doing time series prediction, there’s a custom module in Cortana Intelligence gallery to do time series forecasting.

https://gallery.cortanaintelligence.com/Experiment/Time-Series-Forecasting-using-Custom-Modules-1

You just have to open that in your studio and re-use the built modules in your experiment. See what’s happening to your sales in next December! 🙂

Competing in Kaggle with Azure Machine Learning

MLData science is one of the most trending buzz words in the industry today. Obviously you’ve to have hell a lot of experience with data analytics, understanding on different data science related problems and their solutions to become a good data scientist.

Kaggle (www.kaggle.com) is  a place where you can explore the possibilities of data science, machine learning and related stuff. Kaggle is also known as “the home of data science” because of it’s rich content and the wide community behind it. You can find out hundreds of interesting datasets uploaded by data science enthusiasts all around the world on Kaggle. The most fascinating thing that you can find on Kaggle is competitions! Some competitions are bound with exciting prize tags while some competitions offer wonderful job opportunities when you score a top rank on it.

As we discussed in previous posts, Azure Machine Learning enables you to deploy and test predictive analytics experiments easily. Sometimes you need to not to code a single line to develop a machine learning model. So let’s start our journey on Kaggle with Azure Machine Learning.

01. Sign up for Kaggle – Go to kaggle.com & sign up using your Facebook/Google or LinkedIn account. It’s totally free! 🙂

Kaggle landing page

Kaggle landing page

02. Register for a Kaggle competition – Under the competition section, you can find out many competitions. Will start from a simple experiment that doesn’t go with any prize tag or job offering but worth enough to try out as your first experience on Kaggle.

Can you classify monsters?

Can you classify monsters?

03. Ghouls, Goblins, and Ghosts… Boo! Search for this competition categorized under ‘Knowledge’ sector of the competitions.  The task you have to do in the competition is described precisely on ‘Competition Details’

04. Get the data – After accepting the terms and conditions of Kaggle, you can download the training dataset, test dataset and the sample submission in .csv format. Make sure to take a deep look on features and understand whether you need some kind of data preprocessing before jumping into the task 😉

05. Understand the problem – You can easily figure out this is a multi-class classification machine learning problem. So let’s handle it on that way!

06. Get the data to your Studio – Here comes Azure Machine learning! Go to AML Studio (Setting up Azure Machine Learning is discussed here) and upload the data files through ‘Add Files’ option.

07. Build the classifier experiment – Same as building a normal AML experiment. Here I’ve split the training dataset to evaluate the model. The model with highest accuracy has chosen to do the predictions. ‘Tune model hyperparameter’ has used to find the optimal model parameters.

Classifier Experiment

Classifier Experiment

08. Do the prediction – Now it’s time to use the trained model to predict the type of the ghost using the data in test dataset. You can download the predicted output using ‘Convert to CSV’ module.

Predicting with the trained model

Predicting with the trained model

09. Submission – Make sure to create the output according to the sample submission.

10. Upload the submission to Kaggle –  You can compete as a team or individual. See where you are in the list!

Here's I'm the 278th! :)

Here’s I’m the 278th! 🙂

That’s it! You’ve just completed your first Kaggle competition. This might not lift you to the top of the competitors list. But it’s not impossible to use Azure Machine Learning in real world machine learning related problem solving.