Getting formula for regression line in r

Question

There are a couple of answers on stackoverflow that show how to return a formula for a regression line, but I cannot figure out how to get them to work and give me a formula.

Here is one example that I am trying to use that takes a dataframe.

In my code, I am reading in data from a table:

vafs <- read.table("outputFile", header = TRUE) sample1 <- vafs$Sample1 sample2 <- vafs$Sample2

And then plotting generally like this:

lm_eqn <- function(df,y,x){ m <- lm(y ~ x, df); eq <- substitute(italic(y) == a + b %.% italic(x)*","~~italic(r)^2~"="~r2, list(a = format(coef(m)[1], digits = 2), b = format(coef(m)[2], digits = 2), r2 = format(summary(m)$r.squared, digits = 3))) as.character(as.expression(eq)); } p <- ggplot(vafs, aes(x=sample1, y=sample2, alpha=0.5, label=identity, size=15)) + geom_text(aes(x = 10, y = 300, label = lm_eqn(vafs, sample1, sample2)), parse = TRUE)+ geom_point() + xlim(0,0.003) + ylim(0,0.003) + geom_abline(intercept = 0, slope = 1, size=3)+ # y=x line xlab('\nVAF Individual 1') + ylab('VAF Individual 2\n') + labs(title = 'Muliplier = 1x\n')+ geom_smooth(method=lm, se=TRUE, size=3, colour='red')+ # regression line theme_bw()+ # no gray background theme(panel.border = element_blank())+ # no border theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank())+ # no gridlines theme(axis.title = element_text(size = 50))+ # change label size theme(plot.title = element_text(size = 50))+ # change title size theme(plot.title = element_text(hjust = 0.5))+ # center title theme(axis.text.x = element_text(size = 50, colour="black", angle=90))+ # change tick size theme(axis.text.y = element_text(size = 50, colour="black"))+ # change tick size theme(legend.position="none")+ # no legend theme(axis.ticks = element_line(colour = "black", size = 2))+ # hide ticks theme(axis.line = element_line(colour = "black", size=3)) # add axis jpeg("output2.jpg", units="in", width=17, height=17, res=500) print(p) dev.off()

Then if I try to use the method answered in the link above (also shown below) I can't figure out how to properly pass it my data; just simply passing vafs does not work.

lm_eqn <- function(df){ m <- lm(y ~ x, df); eq <- substitute(italic(y) == a + b %.% italic(x)*","~~italic(r)^2~"="~r2, list(a = format(coef(m)[1], digits = 2), b = format(coef(m)[2], digits = 2), r2 = format(summary(m)$r.squared, digits = 3))) as.character(as.expression(eq)); } p1 <- p + geom_text(x = 25, y = 300, label = lm_eqn(df), parse = TRUE)

Here is some sample input data to understand what i'm taking as input:

Sample1 Sample2 Identity 0.000100576639399 7.01336045166e-05 label 0.000201153278798 0.000263732817381 label 6.70510929328e-05 0.000109685906595 label

missuse · Accepted Answer · 2017-09-26 22:43:23Z

I found it working if I add y, and x arguments to the lm_eqn function:

library(ggplot2) ggplot(cars)+ geom_point(aes(x=speed, y =dist ))+ geom_text(aes(x = 10, y = 300, label = lm_eqn(cars, dist, speed)), parse = TRUE) lm_eqn <- function(df, y, x){ m <- lm(y ~ x, df); eq <- substitute(italic(y) == a + b %.% italic(x)*","~~italic(r)^2~"="~r2, list(a = format(coef(m)[1], digits = 2), b = format(coef(m)[2], digits = 2), r2 = format(summary(m)$r.squared, digits = 3))) as.character(as.expression(eq)); }

EDIT: with annotate I had to define x and y in lm_eqn explicitly since they were not defined in the aes call. However the improvement in the look is worth it:

ggplot(cars)+ geom_point(aes(x=speed, y =dist ))+ annotate(geom = "text",x = 10, y = 300, label = lm_eqn(cars, cars$dist, cars$speed), parse = TRUE)

It's good practice to replace the geom_text(...) with annotate(geom = "text", ...). The former plots one instance per row of of the data, resulting in the bold/blurry text, while the latter only plots one label.
@missuse This actually seems to almost be working, as everything runs and I get a plot out, but there is no formula displayed on the plot. Am i calling the method improperly?
@Brian thanks, added edit with your suggestion. It does look much better. The Nightman (philly rocks!) does it work with data(cars) as in my example or not? I am using ggplot2 2.2.1. In geom_text I could make it work only if x, y and label are in aes. Can you provide a sample of data with dput.
@missuse (Sunny is the best haha) and yeah I can replicate exactly what you have done, so I'm not sure what the difference is. I have added the first few rows of data so you can get the idea of what it looks like.
@The Nightman x = 10, y = 300 in geom_text while xlim(0,0.003) and ylim(0,0.003), fix the geom_text coordinates. Improvement?

Collectives™ on Stack Overflow

Getting formula for regression line in r

1 Answer 1

8 Comments

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

8 Comments

Linked

Related