Predictive analytics uses data along with analysis, statistical, and machine learning techniques to create a predictive model for forecasting future events. The term “predictive analytics” does not describe a particular statistical or machine learning technique, but, rather, the application of a technique to create a quantitative prediction about the future. Frequently, supervised machine learning techniques are used to predict a future value (for example, the temperature tomorrow will be 30°) or to estimate a probability (for example, the probability of rain tomorrow is 75%).
Predictive analytics is often discussed in the context of Big Data as businesses apply algorithms to derive insights from large data sets using Hadoop and MapReduce. Data sources used to create predictive models often include SQL databases, equipment log files, images, video, audio, and sensor data. These predictive models can then be deployed for production use in an IT environment or on an embedded system.
For example: In the energy industry, demand for electricity is commonly forecasted for the next day. This forecast is used by grid operators to ensure that power plant generation can be scheduled to meet the demand. Other applications include computer vision, consumer credit scores, predictive maintenance of machinery and vehicles, retail forecasting, risk management, and healthcare.
Techniques used in predictive analytics often include some of the following:
See also: Parallel Computing Toolbox, Signal Processing Toolbox, Image Processing Toolbox, Statistics and Machine Learning Toolbox, Neural Network Toolbox, MATLAB, Database Toolbox, data analytics, unsupervised learning, predictive modeling, predictive maintenance, prescriptive analytics