Multiple Regression Analysis (Points and Predictor Grids)

Author: O.Conrad (c) 2004
Menu: Spatial and Geostatistics | Regression

Description

Linear regression analysis of point attributes with multiple grids. Details of the regression/correlation analysis will be saved to a table. The regression function is used to create a new grid with regression based values. The multiple regression analysis uses a forward selection procedure.

References

Bahrenberg, G., Giese, E., Nipper, J. (1992): Statistische Methoden in der Geographie 2 - Multivariate Statistik. Stuttgart, 415p.

Parameters

	Name	Type	Identifier	Description	Constraints
Input	Predictors	grid list, input	`PREDICTORS`	-	-
Input	Points	shapes, input	`POINTS`	-	-
Output	Details: Coefficients	table, output, optional	`INFO_COEFF`	-	-
	Details: Model	table, output, optional	`INFO_MODEL`	-	-
	Details: Steps	table, output, optional	`INFO_STEPS`	-	-
	Residuals	shapes, output, optional	`RESIDUALS`	-	-
	Regression	grid, output	`REGRESSION`	regression model applied to predictor grids	-
	Regression with Residual Correction	grid, output, optional	`REGRESCORR`	regression model applied to predictor grids with interpolated residuals added	-
Options	Grid System	grid system	`PARAMETERS_GRID_SYSTEM`	-	-
	Dependent Variable	table field	`ATTRIBUTE`	-	-
	Resampling	choice	`RESAMPLING`	-	Available Choices: [0] Nearest Neighbour [1] Bilinear [2] Cubic 1 [3] Cubic 2 Default: 3
	Include X Coordinate	boolean	`COORD_X`	-	Default: 0
	Include Y Coordinate	boolean	`COORD_Y`	-	Default: 0
	Intercept	boolean	`INTERCEPT`	-	Default: 1
	Method	choice	`METHOD`	-	Available Choices: [0] include all [1] forward [2] backward [3] stepwise Default: 3
	Significance Level	floating point number	`P_VALUE`	Significance level (aka p-value) as threshold for automated predictor selection, given as percentage	Minimum: 0.000000 Maximum: 100.000000 Default: 5.000000
	Cross Validation	choice	`CROSSVAL`	-	Available Choices: [0] none [1] leave one out [2] 2-fold [3] k-fold Default: 0
	Cross Validation Subsamples	integer number	`CROSSVAL_K`	number of subsamples for k-fold cross validation	Minimum: 2 Default: 10
	Residual Interpolation	choice	`RESIDUAL_COR`	-	Available Choices: [0] Multilevel B-Spline Interpolation [1] Inverse Distance Weighted Default: 0

Command Line


Usage: saga_cmd statistics_regression 1 [-PREDICTORS ] [-POINTS ] [-ATTRIBUTE ] [-INFO_COEFF ] [-INFO_MODEL ] [-INFO_STEPS ] [-RESIDUALS ] [-REGRESSION ] [-REGRESCORR ] [-RESAMPLING ] [-COORD_X ] [-COORD_Y ] [-INTERCEPT ] [-METHOD ] [-P_VALUE ] [-CROSSVAL ] [-CROSSVAL_K ] [-RESIDUAL_COR ]
  -PREDICTORS:     Predictors
	grid list, input
  -POINTS:         Points
	shapes, input
  -ATTRIBUTE:      Dependent Variable
	table field
  -INFO_COEFF:     Details: Coefficients
	table, output, optional
  -INFO_MODEL:     Details: Model
	table, output, optional
  -INFO_STEPS:     Details: Steps
	table, output, optional
  -RESIDUALS:      Residuals
	shapes, output, optional
  -REGRESSION:     Regression
	grid, output
  -REGRESCORR:     Regression with Residual Correction
	grid, output, optional
  -RESAMPLING:
                        Resampling
                        	choice
                        	Available Choices:
                        	[0] Nearest Neighbour
                        	[1] Bilinear
                        	[2] Cubic 1
                        	[3] Cubic 2
                        	Default: 3
  -COORD_X:        Include X Coordinate
	boolean
	Default: 0
  -COORD_Y:        Include Y Coordinate
	boolean
	Default: 0
  -INTERCEPT:      Intercept
	boolean
	Default: 1
  -METHOD:
                        Method
                        	choice
                        	Available Choices:
                        	[0] include all
                        	[1] forward
                        	[2] backward
                        	[3] stepwise
                        	Default: 3
  -P_VALUE:
                        Significance Level
                        	floating point number
                        	Minimum: 0.000000
                        	Maximum: 100.000000
                        	Default: 5.000000
  -CROSSVAL:
                        Cross Validation
                        	choice
                        	Available Choices:
                        	[0] none
                        	[1] leave one out
                        	[2] 2-fold
                        	[3] k-fold
                        	Default: 0
  -CROSSVAL_K:     Cross Validation Subsamples
	integer number
	Minimum: 2
	Default: 10
  -RESIDUAL_COR:
                        Residual Interpolation
                        	choice
                        	Available Choices:
                        	[0] Multilevel B-Spline Interpolation
                        	[1] Inverse Distance Weighted
                        	Default: 0