Machine Learning API’s for Retail

One of the interesting announcements that came out of Strata+Hadoop World last week was the new machine learning (ML) capabilities in the Azure Marketplace.

Not only can data scientists build models and publish it as a webservice, but they can also share it with the rest of the world via the Azure Marketplace. The Azure Marketplace enables data scientists to share and monetize their ML API’s.

Retailers can search for published API’s, subscribe to them and use them in their applications. The Azure Marketplace has some existing web services that are great starters :

Many of these are Retail specific:

Market Basket Analysis:
This is an API built with Azure Machine Learning that a helps your customers discover items in your catalog that are frequently purchased together. Use your customer purchase history to add "Frequently Bought Together" recommendations to your website and to improve conversion in your digital store.

Sentiment Analysis: 

This web service can be used to classify text into positive, negative and neutral text. A score is also provided to indicate the intensity level. A score of 0 indicates a neutral sentiment and any value above 0 indicates a positive sentiment while any negative value indicates a negative sentiment (score ranges between -1 and 1). The dictionary used within the web service is based on MPQA, a commonly used polarity dictionary, and the web service is built to deal with negation.

Predictive Maintenance:

Survival Analysis API is an example built with Microsoft Azure Machine Learning that computes the probability that an event occurs by a certain point in time. Under many scenarios, the main outcome under assessment is the time to an event of interest. In other words, it addresses the question “when will this event occur”. Examples include situations where the data describes the elapsed time (days, years, mileage, etc.) until the event of interest (disease relapse, PhD degree received, brake pad failed) occurs. Each instance in the data represents a specific object (a patient, a person, a car, etc.).

Inventory/Sales Forecasting:

This API is an example built with Microsoft Azure Machine Learning that fits an ARIMA model to user inputted data and outputs predicted forecasted value for future dates. Will the demand for a specific product increase this year? Can I predict my product sales for the Christmas season, so that I can effectively plan my inventory? Forecasting models are apt to address such questions. Given the past data, these models examine hidden trends and seasonality to predict future trends.


Recommendations API is an example built with Microsoft Azure Machine Learning that helps your customer discover items in your catalog. Customer activity in your digital store is used to recommend items and to improve conversion in your digital store. The recommendation engine may be trained by uploading data about past customer activity or by collecting data directly from your digital store. When the customer returns to your store you will be able to feature recommended items from your catalog that may increase your conversion rate. Microsoft Azure Machine Learning’s Recommendations includes Item to Item recommendations: a customer who bought this also bought that and Customer to Item recommendations: a customer like you also bought that.

You can find details of the announcement over at the Machine Learning Blog