Analysis Tools Menu: Difference between revisions

From Eigenvector Research Documentation Wiki
Jump to navigation Jump to search
imported>Scott
 
(9 intermediate revisions by 2 users not shown)
Line 1: Line 1:
__TOC__
__TOC__
==Preprocessing==
The [[Analysis GUI]] "Tools" menu contains a number of tools which can be used to improve or assess model stability or validity as well as investigate general data information content.


Whithin the Analysis GUI, the Preprocess menu allows setting and viewing of preprocessing options. Each block has a submenu for setting common preprocessing methods directly or for accessing the Custom Preprocessing GUI. After making a change to preprocessing, you will usually be required to recalculate the model. '''Preprocess>Plot Preprocessed Data''' menu allows you to displays the effects of the selected preprocessing on loaded data. The plot(s) are automatically updated to reflect any changes in preprocessing.
==Cross Validation==
 
[[Image:Preprocessing.png|Preprocessing GUI ]]
 
Custom Preprocessing window after '''Show''' button has been clicked.


The Custom Preprocessing window allows you to choose from a list of available preprocessing methods. After selecting a method from the available methods list, pressing the Add button will insert the selected method into the Selected Methods list on the right. Some methods will display an additional window for method specific settings when the item is added to the Selected Methods list. These settings can also be modified by selecting the method in the Selected Methods window, then pressing the Settings… button. Clicking '''OK''' load the new preprocessing structure into the GUI and the Status Pane will indicate that the model will need to be recalculated.
Cross-Validation displays an interface for selecting and configuring a cross-validation method (see [[Using Cross-Validation]] for more information). Cross-Validation is not available for all methods.
 
The Selected Methods are order specific – that is, they get applied in order from top to bottom. The order of preprocessing can have a very significant effect on an analysis; this should be carefully considered in the development of a model.
 
Note that n-way data have different preprocessing methods than 2-way data, and care should be taken when switching between 2-way and n-way data. Again, you must recalculate the model after making any changes to preprocessing.
 
You can load and save preprocessing settings from/to the MATLAB workspace and .mat files using the Custom Preprocessing window's '''Load''' button.
The '''Show''' button will display two plots side by side with the raw data on the left and the preprocessed data on the right. You can select the axis values using the drop-down menus and toggle the Excluded button to show excluded data.
==Cross Validation==


Cross-Validation displays an interface for selecting and configuring a cross-validation method (see [[crossval]] in the reference manual for more information). Cross-Validation is not available for all methods.
[[Image:Crossval_gui.jpg|Cross Validation GUI ]]
[[Image:Crossvalgui.png|Cross Validation GUI ]]


Cross-Validation GUI.
Cross-Validation GUI.
Line 36: Line 22:
==Estimate Factor SNR (Signal to Noise Ratio)==
==Estimate Factor SNR (Signal to Noise Ratio)==


Displays a plot of an estimate of the signal to noize ratio for each principal
Displays a plot of an estimate of the signal to noise ratio for each principal
component. Ratios of 2 or below are dominated by noise, above 3 are OK,
component. Ratios of 2 or below are dominated by noise, above 3 are OK,
and between 2 and 3 are a judgment call. The number of factors needed to
and between 2 and 3 are considered threshold and their use is a judgment call. The number of factors needed to
describe the data is the number of eigenvectors with signal to noise
describe the data is the number of eigenvectors with signal to noise
ratios greater than about 2.
ratios greater than about 2.
Line 45: Line 31:
[[Image:Estimatefactorsnr.png|Estimate Factor SNR Plot ]]
[[Image:Estimatefactorsnr.png|Estimate Factor SNR Plot ]]


Estimate Factor SNR Plot.
Estimate Factor Signal to Noise Ratio Plot.


See [[estimatefactors]] for more information.
See [[estimatefactors]] for more information.
Line 51: Line 37:
==Test Model Robustness==
==Test Model Robustness==


The '''Tools>Test Model Robustness''' menu displays sections for testing a regression model for robustness to various effects.
The '''Tools>Test Model Robustness''' menu displays sections for testing a regression model for robustness to various effects using the [[testrobustness]] function.


Given a regression model and validation (test) data x and y, this function performs
Given a regression model and validation (test) data x and y, this function performs
Line 58: Line 44:
mapped as a function of the test parameters.
mapped as a function of the test parameters.


The two types of tests performed:
The three types of tests performed:
* '''interference''' : adds a gaussian peak of varying width and position tothe x-block data. Results are mapped by peak width and position.
* '''interference''' : adds a gaussian peak of varying width and position tothe x-block data. Results are mapped by peak width and position.
* '''shift''' : shifts and broadens the x-block data and maps the results relative to shift amount and width of broadening filter. Filter is based on savgol filter of a specified order and width (see options.)
* '''shift''' : shifts and broadens the x-block data and maps the results relative to shift amount and width of broadening filter. Filter is based on savgol filter of a specified order and width (see options.)
* '''singlevar''' : adds the standard deviation of each variable and divides it by 100. This loop is done one variable at a time. Results are differences in prediction between the original samples and the perturbed samples.

Latest revision as of 07:02, 6 December 2023

The Analysis GUI "Tools" menu contains a number of tools which can be used to improve or assess model stability or validity as well as investigate general data information content.

Cross Validation

Cross-Validation displays an interface for selecting and configuring a cross-validation method (see Using Cross-Validation for more information). Cross-Validation is not available for all methods.

Cross Validation GUI

Cross-Validation GUI.

The Apply button will change crossval settings in Analysis but not close the GUI. Clicking OK will change the crossval setting and close the GUI. Reset will changes the setting in the crossval GUI to the original values and Cancel closes the GUI without making changes.

Correlation Map

The Tools/Correlation Map menu you can select 3 different options for producing a pseudocolor map which shows the correlation of between variables (columns) in a data set.

  • No variable reordering (keep original order).
  • Standard sign-sensitive variable reordering.
  • Absolute value variable reordering (positive or negative correlated variables are grouped together).

Estimate Factor SNR (Signal to Noise Ratio)

Displays a plot of an estimate of the signal to noise ratio for each principal component. Ratios of 2 or below are dominated by noise, above 3 are OK, and between 2 and 3 are considered threshold and their use is a judgment call. The number of factors needed to describe the data is the number of eigenvectors with signal to noise ratios greater than about 2.


Estimate Factor SNR Plot

Estimate Factor Signal to Noise Ratio Plot.

See estimatefactors for more information.

Test Model Robustness

The Tools>Test Model Robustness menu displays sections for testing a regression model for robustness to various effects using the testrobustness function.

Given a regression model and validation (test) data x and y, this function performs one of several tests to assess the effect of different perturbations on the test data x-block. The effect on prediction error, Q and T^2 are mapped as a function of the test parameters.

The three types of tests performed:

  • interference : adds a gaussian peak of varying width and position tothe x-block data. Results are mapped by peak width and position.
  • shift : shifts and broadens the x-block data and maps the results relative to shift amount and width of broadening filter. Filter is based on savgol filter of a specified order and width (see options.)
  • singlevar : adds the standard deviation of each variable and divides it by 100. This loop is done one variable at a time. Results are differences in prediction between the original samples and the perturbed samples.
Retrieved from "https://www.wiki.eigenvector.com/index.php?title=Analysis_Tools_Menu&oldid=12070"