Cooksd: Difference between revisions
Jump to navigation
Jump to search
imported>Benjamin |
imported>Benjamin mNo edit summary |
||
| Line 1: | Line 1: | ||
===Purpose=== | ===Purpose=== | ||
Calculates Cooks Distance | Calculates Cooks Distance for samples in a regression model. | ||
===Synopsis=== | ===Synopsis=== | ||
| Line 8: | Line 8: | ||
===Description=== | ===Description=== | ||
Cooks distance measures how much the model would change if a given sample is left out. Specifically, how much all of the fitted values change when the ith sample is removed. A larger distance value indicates a sample has a stronger influence on the fitted values. | Cooks distance measures how much the model would change if a given sample is left out. Specifically, Cooks distance is a metric of how much all of the fitted values change when the ith sample is removed. A larger distance value indicates a sample has a stronger influence on the fitted values. | ||
* Distance > 0.5: ith sample may be influential and may be worthy of further investigation. | * Distance > 0.5: ith sample may be influential and may be worthy of further investigation. | ||
| Line 15: | Line 15: | ||
====Inputs==== | ====Inputs==== | ||
* '''model''' = | * '''model''' = A standard model structure. | ||
('''Note:''' Currently only PLS models are supported.) | ('''Note:''' Currently only PLS models are supported.) | ||
====Outputs==== | ====Outputs==== | ||
* '''distances''' = | * '''distances''' = An m-by-1 vector of the calculated Cooks distances for each sample. | ||
===See Also=== | ===See Also=== | ||
[[plotscores]], [[leverag]], [[figmerit]], [[pls]] | [[plotscores]], [[leverag]], [[figmerit]], [[pls]] | ||
Latest revision as of 12:30, 16 August 2017
Purpose
Calculates Cooks Distance for samples in a regression model.
Synopsis
- distances = plotscores(model)
Description
Cooks distance measures how much the model would change if a given sample is left out. Specifically, Cooks distance is a metric of how much all of the fitted values change when the ith sample is removed. A larger distance value indicates a sample has a stronger influence on the fitted values.
- Distance > 0.5: ith sample may be influential and may be worthy of further investigation.
- Distance > 1: ith sample is quite likely to be influential.
Inputs
- model = A standard model structure.
(Note: Currently only PLS models are supported.)
Outputs
- distances = An m-by-1 vector of the calculated Cooks distances for each sample.