Making predictive applications has never been so easy and cheap as now. Although we do not realize, applications with predictive capabilities coexist with us for a long time: spam detectors (predict whether a new mail is sent to the spam folder based on certain patterns), profiles recommendations on Twitter (predicts which users can interest us based on patterns of other users that look like us), Google ads (predict which ads are more likely to click), or your phone company (predict if you’re going to unsubscribe based on your consumption patterns).
Large companies have their own data centers, analysts, scientists, mathematicians, programmers and, above all, budget to deal with this type of projects. But since a few years ago, cloud services are appearing in the market that bring predictive technologies closer to companies of any size.
It’s the low-cost Machine Learning.
The last company to sign up has been Amazon, a company that has been using Machine Learning internally for years to recommend its products based on visitor behavior on the web. If you are an Amazon customer, you will hear “What other products do customers buy after seeing this product?” Or “Customers who bought this product also bought …”.
Amazon launches its low-cost Machine Learning service
Amazon has opened the doors to a core technology in its success as a company. The company announced his Machine Learning service to help companies “to use all the data they have collected to improve the quality of their decisions”. Yes, it is not only about offering a better service to customers, but about making business decisions based on data. Predicting which is the best place to open a new store in a big city is not trivial. Predicting the nationalities of guests in a hotel chain serves to improve the management of resources (food, staff, activities …). Disgruntled customers usually leave without notice. Would you like to anticipate and recover it even before he has made the decision?
It is not that Amazon has suffered an attack of generosity to deliver to the market his hen of the golden eggs. What it is really doing is positioning itself in a market, that of low-cost predictive engines, which has been developing for a few years.
Indeed, in 2010 Google launched its Prediction API, an environment that allowed programmers to make “smart” applications that learn from data.
A year later, BigML launched its machine learning cloud service that brings this technology to non-programmers, with really low costs (even free), enriching visualizations and a really impressive ease of use.
Just two months ago Microsoft made the official announcement of its own Machine Learning, an environment that runs on the Azure infrastructure.
These are cloud environments, flexible and cheap. We are sure that in your company you have a lot of data. We know you’ve always used them. But … have you thought about including the new types of data that are generated in the digital environment? Emails with your clients, visits to the web, requests for information on forms, downloads of documentation, interactions on social networks, data on purchases with credit cards, phone calls … That’s your Big Data. If you want to take advantage and add competitive advantages, it’s time to use the same tools as the big ones.