Loading

ESDTEST

gts outlier

The ESDTEST function detects outliers in a GTS (or a LIST of GTS), by applying a generalized extreme studentized deviate test.

This test is done under the assumption that the GTS follows an approximately normal distribution.

A Grubbs’test is done for one candidate at a time. Then, the candidate is removed from the set and another Grubbs’test is performed. This process is iterated a given number of times. The detected outliers are the removed values and the current candidate of the last successful test.

A LIST of ticks (or a LIST of LIST of ticks), corresponding to the outliers, is pushed back onto the stack.

This function only applies to GTS of type DOUBLE.

Rosner, Bernard (May 1983), “Percentage Points for a Generalized ESD Many-Outlier Procedure”,Technometrics, 25(2), pp. 165-172.

ESDTEST is available since version 1.0.0.

See also

Signatures

Examples

// Macro used to generate an approximately normal distribution using central limit theorem <% RAND RAND RAND RAND RAND RAND + + + + + 3.0 - %> 'normal' STORE // GTS with an approximately normal distribution NEWGTS 1 1000 <% NaN NaN NaN @normal ADDVALUE %> FOR // we add outliers (> 3.0 in absolute value) 368 NaN NaN NaN -3.1 ADDVALUE 422 NaN NaN NaN 3.0001 ADDVALUE 456 NaN NaN NaN 9.8 ADDVALUE 643 NaN NaN NaN -200.9 ADDVALUE DEDUP 4 F ESDTEST