K-Nearest Neighbour
A technique that classifies each record in a data set based on a combination of the classes of the K record(s) most similar to it in a historical data set.

K-S Statistic
Acronym for Kolmogorov-Smirnoff Statistic.

KBS
Abbreviation for Knowledge Based Systems. A computer program that uses a knowledge base and inference procedures to solve problems. These systems do not use databases.

KDD
Abbreviation for Knowledge Discovery. KDD involves the following steps: data preparation, exploratory data analysis, descriptive or predictive modelling, performance and value assessment, and strategy development.

Key
A value that is used with a crypographic algorithm to encrypt. The longer the key, the more secure the encryption.

Key Name
In a decision tree, the value assigned to a decision key that labels every node.

Key Value
In a decision tree, the value assigned to a decision key. Key values are listed in each node.

Key Value Break Points
In a decision tree, a set of Key Values that lead to set of branches.

Keyline
This is a combination of numbers and letters, usually beginning with the postal code, which is used as a rough household duplicate eliminator. Also see Match Key.

Keypunching
The process whereby names and addresses and other data from hard copy are entered onto a computer tape or disk. It is performed on a typewriter-like keyboard. See Data Entry.

Kill File
A file that tells your newsreader which newsgroup articles you always want to skip.

Kitchen Sinks
Mortgage-backed securities composed of the scraps and pieces of other mortgage-backed securities.

Kiting
A scheme where a merchant submits a sales transaction on his/her personal bankcard account(s) to obtain cash advances through their business.

Knowledge Discovery (KDD)
Knowledge discovery involves the following steps: data preparation, exploratory data analysis, descriptive or predictive modelling, performance and value assessment, and strategy development. Use of the term knowledge discovery implies that the patterns discovered must be valid, useful, understandable, and previously unknown within the collection of data.

Knowledge-Based Systems (KBS)
A computer program that uses a knowledge base and inference procedures to solve problems. These systems do not use databases.

Kolmogorv-Smirnoff Statistic (K-S Statistic)
A statistic identifying the score where a scoring model is most powerful. See Scorecard Power.

Kurtosis
Characterises relative peakedness or flatness of a given distribution compared to a normal distribution. It is the fourth moment of a distribution. Since the unconditional normal distribution has a kurtosis of 3, excess kurtosis is defined as Kx-3. See Leptokurtosis, Excess Kurtosis and Fat Tails.