Not in theory, in practice.
The number scenarios where machine learning is used today to improve services is nearly unlimited. Here is just a small list:
Prediction of stock and sales.
Using its own stock and sales data from the past together with data from their web logs, marketing campaigns as well as external data such as weather data, currency data and market fluctuations a shop builds a model that allows it to make much better predictions for sales and stock over the next 6 months.
Mail categorization and routing.
Models trained on past data and with automated translation allow companies to effectively route incoming letters and mail traffic to the right company division without the need for manual reading and assignment.
Transportation and time of arrival.
Using various internal and external factors, transportation companies are able improve considerably on the estimated delays and times of arrival, leading to better service and reduction of cost for waiting times.
Visual product and part recognition.
Using a library of images of products and parts, a company builds an application that allows field engineers to get all relevant data on a product or a part by simply taking a picture, which then is analysed and matched with the library.
Recommendation system for online customers.
Based on a customers searches and orders on a website or portal, combined with a knowledge about the assets that the customer has acquired in the past, a system produces recommendations tailored to the user’s personal preferences and situation.
These are 5 out of 5 trillion ideas…
… out there to be realized. If it’s about data, about classification, recognition, prediction, annotation or optimization, there is likely to be a use case for machine learning not far away. Let yours be the next!