Helps you divide your customers into different homogenous groups or segments, and used for defining marketing or risk strategies for each segment

These segments are often “discovered” from the data, using techniques such as clustering

The most widely used purpose of analytic models. For predicting customer behaviour (such as retention, defaults, response to a marketing offer, etc.), and develop interventions based on the same

Involves techniques such as decision trees, logistic regression, neural networks, etc.

Used to project the future business metric based on historical data. For example, Sales forecasting by SKU’s and store (sales), or loss forecasting for a loan portfolio (risk management)

More complex, typically involving handling Longitudinal data, using Time Series, Regression or a combination of both

Explanatory models are developed to discover complex relationships between various phenomena.

For instance, a risk manager at a consumer lending institution would like to understand the underlying drivers of their customers' ability and willingness to pay. The factors could include demographics, past behaviour, econometric factors, product characteristics, etc.

The output of an explanatory model is a set of graphs, charts or tables that clearly help managers identify key drivers and validate or disprove hypotheses

Predictive models are developed to classify customers, accounts or prospects or predict their behaviour

For instance, a lending institution could use a predictive model (or a series of models) to estimate the propensity of an applicant or a customer to default on a loan, at the time of originations, and at different points in the course of the relationship

The output from a predictive model is a set of one or more scores that is used for decision making in the business process, or fed into a decision model

Decision models help identify the best possible decision in a given situation

For instance, a direct marketer in a retail financial institution would like to make decisions regarding: who to target (the target customers), what offer to make (the products and features), what channels (telephone, direct mail, email), what message to use (creatives). These decisions need to be made within time, budgetary constraints and other considerations. The direct marketer is likely to use optimization techniques and simulators to identify the right set of decisions to make

The output from the decision models help identify the outcomes associated with certain interventions that are being planned by the decision maker