Classification is the grouping of a data set, based on some predefined criteria. The criteria are usually based on some historic information, and classification tries to classify the data set, based on information received from that historic criteria. An example: A company wants to have a database of 1 million customers in the United States, including their demographic information. It wants to identify the top 50,000 customers who have highest propensity to respond to an offer campaign. The company’s analyst retrieves past data on response rates for a similar campaign on 200,000 customers. Their response rate is trained on a classification technique that tries to separate respondents with nonrespondents and also create a scorecard for the customers. The model is then executed on a 1-million-customer base, to classify respondents from nonrespondents and pick the top 50,000 respondents who should be sent the new campaign.
Other examples include
- Google identifying whether a mail is spam, based on its content and other information
- Assessing whether an employee would attrite, based on his/her past information
