MACHINE LEARNING FOR HOUSEHOLDER
machine learning for householder
Machine learning - Trend
A long time ago the course is Andrew Ng, who explains everything in simple terms so that a rather complicated material does not understand even the most diligent student. Since then, the topic of machine learning has become close to me, and I periodically look at projects in the area of Big Data (read the previous column), and in the field of machine learning.
In addition to the large number of startups that somewhere inside themselves using machine learning algorithms that are already available several services that offer machine learning as a service! That is, they provide the API, which you can use in your projects, with no delving into how the analysis is carried out and data prediction.
Google Prediction API
One of the first to offer Machine Leaning as a Service became Google! For quite a long time, anyone can use the Google Prediction API (literally «API for predictions"). Up to a certain amount of data you can use it absolutely free, just having got an account on the Google In game Prediction the API . What kind of predictions do you mean? The task can be different: to determine the future value of a parameter on the basis of available data or determine the identity of an object to some of the types (eg, text language: Russian, French, English).
After registration you receive full access to RESTful API, based on which one can build, say, a reference system, detection of spam and suspicious activity, analyze user behavior and more. Already we managed to appear interesting projects built on the basis of intensive use Google Prediction API, for example Pondera Solutions, which uses machine learning to build on Google's antifraud systems.
As an experiment, you can take a ready-made data model: the language identifier for the construction of the system, determine which language is written incoming text or mood identifiers to automatically determine the tone of the comments that users leave. I think in the future we will talk about Google Prediction API in detail.
BigML is a leading Machine Learning company that helps other businesses make highly automated, data-driven decisions. BigML has pioneered the Machine Learning. BigML platform can be used to analyze and predict: ✓ customer behavior, to increase customer loyalty ✓ site visit behavior ✓diagnostics, to support in health care, hardware maintenance and many other area's ✓risk profiles, to process loan applications ✓stock levels, to optimize supply of goods and so on.
BigML is good for analysis data, so I recommend to use BigML in machine learning.
All operations are carried out in an understandable admin panel (I will not describe the nuances, all will be extremely accessible).
- Choose a CSV-file, which stores the lines describing the characteristics of different kinds of flowers as a source of data (Source).
- Then use this data to build a data set (Dataset), indicating that the need to predict the type of flower. BigML automatically parse the file and, after analysis, to build various graphs visualizing data.
- On the basis of the DataSet is constructed with a single click model on which to base predictions. And BigML again renders the model, explaining the logic of its work. You can even export the result as the script for Python or any other language.
- Once the model is ready, it is possible to make predictions (Predictions). And to do it in different ways: once set all parameters of a flower, or to answer questions of the system, which is based on the situation, is to ask just what she needed.
The same thing could crank out and without UI, and communicating with BigML through BigMLer console application or through a REST API, communicating from the console curl'om usual.
nside BigML and Google Prediction API nothing extraordinary there. And intelligent developers can implement similar engines on their own, so as not to pay third-party services. services can just give a person a night to understand that machine learning - it is not only cool and trend, but also quite simple in many situations. And they can be used to quickly sketch a prototype of the new features for your application or service and test the idea with little or no effort