Please use this glossary to familiarise yourself with some of the more technical terms we use on the site
- Algorithm
A series of business rules written in programming code to run a computational process - such as allocating categories or probabilities to a customer file.
- Categorical Variable
A variable containing absolute values. Examples include Gender, State & Marital Status.
- CHAID
A decision tree approach to segmentation.
- Clustering
Clustering is a statistical technique used to classify objects in to different groups based on some similarity or similarities
- Continuous Variable
A variable containing values on a continuous scale. Examples include Age, Income & Height.
- Control Group
A group of individuals or entities that are isolated from a testing environment and can therefore be used as a benchmark.
- Correlation
Measure of the “strength” of a relationship between two variables i.e. the level to which variables are related.
- Dependent variable
The event or value that one attempts to predict when using predictive modelling.
- Dummy Variable
The creation of a new variable to represent the value from another variable. For example, dummy variables to represent Male & Female could be created from a Gender variable.
- ETL
The Extraction, Transformation and Loading of data.
- Gains Chart
A graphical representation demonstrating the performance of a predictive model.
- Geodemographics
The classification of a population into groups based on geographic and demographic characteristics.
- Independent Variable
The information that one uses to help predict an event or value.
- LTV
Lifetime Value - The calculation of a customer's value to an organisation over their 'lifetime'.
- NPV
Net Present Value - The process of calculating the value of money at a point in time.
- Predictive modelling
Statical process to predict the future behaviour of a customer
- Rank
To place records in order based on a value or set of values.
- RFM
A segmentation approach driven by Recency, Frequency & Monetary Value characteristics.
- Scorecard
A predictive model outputs a scorecard which asigns a probability score to a customer record indicating their likelihood to behave in certain way e.g. likelihood to churn
- Segmentation
The separation of individuals or entities into groups that share similar characteristics or needs.
- Statistically Significant
The process of calculating the likelihood that a result can be attributed to something other than chance.
- Time Series
The analysis of data over a period of time.
