Not only does Azure ML provide a predictive model building tool, but it also offers a service which may be used to deploy solutions. What is more, it makes it possible to easily launch ready-for-use web services which may then be used by desktop- or mobile-type applications.
Predictive analytics is based on algebraic and statistical theories which, drawing on historical data, allow predicting future trends and events. Azure Machine Learning is particularly efficient in this regard, because it makes use of a ready algorithm library (Cortana Intelligence Gallery) or modifies and combines them appropriately with the help of a user-friendly web interface. In case of Azure ML the business value presents itself in no time.
Azure Machine Learning has everything needed to build a cloud predictive solution, from algorithm library through environment to model building and easy model deployment in the form of a ready-for-use web service. Azure ML is built so as to not impose any architecture. It operates irrespective of existing IT infrastructure and it is technologically agnostic – it integrates with libraries written in Python or R.
Example uses of Azure Machine Learning:
- Fraud detection detecting suspicious financial transactions, spam filtering
- Sound-based speech recognition (Speech API)
- Recommendations recommending content suited to particular users, predictive content loading, improving UX applications
- Predicting market needs predicting needs e.g. for goods available in a particular region, for processing power in a region, for energy resources
- Targeted marketing allocating appropriate advertisements to users, choosing appropriate marketing campaigns based on user profiles, cross-selling, up-selling
- Content classification categorizing documents, allocating documents to individuals (e.g. connecting candidate CVs to HR departments)
- Resignation forecasting finding subscribers who may unsubscribe, finding clients who may switch from a free plan to a paid subscription
- Customer service predictive forwarding of messages from clients (e.g. directly to an appropriate department based on message content), social media analysis.
It can also be used for less commercial purposes, e.g.:
- Image-based face recognition (Face API)
- Decoding brain signals, analyzing neuropsychological data
- Car telemetry
- Plane and train delay
- Estimating the risk of disease incidence (e.g. cancer, heart disease)
- Image- and handwriting-based character recognition