VN Multivariate Regression (INTERNAL CALL FOR VN.reg)

Share:

Description

Called by VN.reg for multivariate regression analysis.

Usage

1
2
3
VN.M.reg(B, y, order = NULL, s.t.n = 0.9, n.best = 1, type = NULL,
  point.est = NULL, plot = FALSE, residual.plot = TRUE, location = NULL,
  precision = "LOW", text = FALSE, noise.reduction = FALSE, norm = NULL)

Arguments

B

Complete dataset of independent variables (IV) in matrix form.

y

Dependent variable (DV).

order

Controls the number of the VN.reg.

s.t.n

Signal to noise parameter, sets the threshold of VN.dep which reduces "order" when order=NULL. Defaults to 0.9 to ensure high dependence for higher "order" and endpoint determination.

n.best

Sets the number of closest regression points to use in kernel weighting. Defaults to 1. Should be validated on hold-out set in conjunction with "precision" parameter.

type

Controls the partitioning in VN.reg. Defaults to type="XONLY" for IV based partitioning. type=NULL for both IV and DV partitioning.

point.est

Generates a fitted value of y for a vector or matrix of IV coordinates.

plot

Generates a 3d scatter plot with regression points using plot3d

residual.plot

Generates a matplot for Y.hat and Y

location

Sets the location of the legend

precision

Increases speed of computation at the expense of precision. 2 settings offered: "LOW" (Default setting), and "HIGH". "HIGH" is the limit condition of every observation as a regression point.

text

If performing a text classification, set text=TRUE. Defaults to FALSE.

noise.reduction

In low signal:noise situations,noise.reduction="mean" uses means for VN.dep restricted partitions, noise.reduction="median" uses medians instead of means for VN.dep restricted partitions, while noise.reduction="mode" uses modes instead of means for VN.dep restricted partitions. noise.reduction=NULL (Default setting) allows for maximum possible fit and specific order specification.

norm

Normalizes regressors between 0 and 1 for multivariate regression when set to norm="std", or normalizes regressors according to VN.norm when set to norm="VN". Defaults to NULL.

Value

Returns the values: "Fitted" for only the fitted values of the DV; "regression.points" provides the points for each IV used in the regression equation for the given order of partitions; "rhs.partitions" returns the partition points for each IV; "partition" returns the DV, quadrant assigned to the observation and fitted value, and "Point.est" for predicted values.

Author(s)

Fred Viole, OVVO Financial Systems

References

Viole, F. and Nawrocki, D. (2013) "Nonlinear Nonparametric Statistics: Using Partial Moments" http://amzn.com/1490523995