scatter.plot: A regression function for simple regression and break point...

In kevinlzheng/RegR: Collective Functions to Perform Generalized Linear Model Graphs

Description

This function performed basic regression analysis with the option to add confidence intervals and prediction intervals; alternative regression approaches, e.g., segmented regression, change point analysis (through rpart) and confidence bounds, quantile regression, and lowess regression splines. This version also added a capacity for transforming percent values (where 0 is probably the minimum) by adding 1 or 100 before log transformation if users choose so. Any x variable with 0s and log is true, the add 1 to the original or 100 times of the original.

Usage

scatter.plot(data = NULL, xvar, yvar, xName = "x", yName = "y",
  xlim = NULL, ylim = NULL, sign = FALSE, xlab = names(data[xvar]),
  ylab = names(data[yvar]), addaxes = TRUE, cex.eq = 1.5, bty = "o",
  type = "p", pch = 1, col = 1, main = "", cex.main = 1,
  cex.lab = 1, cex = 1, cex.axis = 1, lty = 1, las = 1, bg = "gray",
  side = 4, tx.col = 2, xlog.scale = 10, ylog.scale = 10,
  lab.pos = "topleft", labs = "", x.add = 0, add.fit = "none",
  interval = "none", ln.col = "red", level = 0.95, add.cor = FALSE,
  add.reg = FALSE, add.rq = FALSE, nl.rq = FALSE, rq.tau = 0.5,
  log = "", add = FALSE, add.r2 = FALSE, add.interp = FALSE,
  pred.x = NA, pred.y = NA, rounder1 = 2, rounder2 = 0, ...)

Arguments

data	A input data frame, default to be NULL,
xvar	x variable, either column name or index value
yvar	y variable, either column name or index value
xName	xName for making equation labels
yName	yName for making equation labels
xlim	xlim
ylim	ylim
sign	if TRUE then use signif for equation otherwise, use round
xlab	xlabel
ylab	ylabel
addaxes	to plot axes
bty	add a box frame
pch	point type
cex	magnifier
cex.axis	# magnifier
lty	line type
las	direct
bg	background col
side	which side to add
tx.col	is for col of labs ln.col for fitted line color
xlog.scale	either 10 or exp
ylog.scale	either 10 or exp
lab.pos	control position of labels "bottomright" "bottomleft", "topright", "topleft"
labs	add labels
x.add	1 to convert percent x value = 0 to 1+ x value , or 100, then *100+1,
add.fit	either "linear", "lowess", "loess", "segment" (segmented regression), "change" (regression tree)
level	CI level
add.cor	add spearman correlation,
add.reg	will add regression model
add.rq	# add regression quantile
rq.tau	a vector to indicate which quantiles we want to use
log	will control x,y axis if log,
add	if TRUE, then add to existing graph
add.r2	a boolean to indicate if r2
add.interp	will add a interpolation for inverse prediction of x based on pred.y
pred.x	used to predict x value
pred.y	used to predict y
rounder1	round number of digits for x value
rounder2	round number of digits for y value
yvar	for column index of x, y values in data data frame, or column names
inteval	"confidence", "prediction", "none"
rq	choose if we want to do quantile regression
nlrq	only works for logistic fit
addaxes	if true, add axes
ex.eq	equation
x.add	add x

Value

Returns . sum((x-xmean)^2)) = sum(x^2)- (sum(x))^2/length(x)

Examples

set.seed(1)
n = 50
x <- rlnorm(n) + 1:n
x <- (x - min(x))/max(x)
y <- n + 1:n + (rnorm(n))*0.2*n
par(mfrow=c(1,2))
plot(y~x)
scatter.plot(data = data.frame(x,y), xvar= "x", yvar = "y", add.fit = "linear", xlim = c(0,n), 
    ylim = range(y), log = "x", x.add =100)

kevinlzheng/RegR documentation built on May 20, 2019, 9:07 a.m.