I need to create some gam plots in ggplot. I can do them with the general plot function, but am unsure how to do with ggplot. Here is my code and plots with the regular plot function. I'm using the College data set from the ISLR package.
train.2 <- sample(dim(College)[1],2*dim(College)[1]/3)
train.college <- College[train.2,]
test.college <- College[-train.2,]
gam.college <- gam(Outstate~Private+s(Room.Board)+s(Personal)+s(PhD)+s(perc.alumni)+s(Expend)+s(Grad.Rate), data=train.college)
par(mfrow=c(2,2))
plot(gam.college, se=TRUE,col="blue")
See update below old answer.
Old answer:
There is an implementation of GAM plotting using ggplot2 in voxel library. Here is how you would go about it:
library(ISLR)
library(mgcv)
library(voxel)
library(tidyverse)
library(gridExtra)
data(College)
set.seed(1)
train.2 <- sample(dim(College)[1],2*dim(College)[1]/3)
train.college <- College[train.2,]
test.college <- College[-train.2,]
gam.college <- gam(Outstate~Private+s(Room.Board)+s(Personal)+s(PhD)+s(perc.alumni)+s(Expend)+s(Grad.Rate), data=train.college)
vars <- c("Room.Board", "Personal", "PhD", "perc.alumni","Expend", "Grad.Rate")
map(vars, function(x){
p <- plotGAM(gam.college, smooth.cov = x) #plot customization goes here
g <- ggplotGrob(p)
}) %>%
{grid.arrange(grobs = (.), ncol = 2, nrow = 3)}
after a bunch of errors: In plotGAM(gam.college, smooth.cov = x) :
There are one or more factors in the model fit, please consider plotting by group since plot might be unprecise
To compare to the plot.gam:
par(mfrow=c(2,3))
plot(gam.college, se=TRUE,col="blue")
You might also want to plot the observed values:
map(vars, function(x){
p <- plotGAM(gam.college, smooth.cov = x) +
geom_point(data = train.college, aes_string(y = "Outstate", x = x ), alpha = 0.2) +
geom_rug(data = train.college, aes_string(y = "Outstate", x = x ), alpha = 0.2)
g <- ggplotGrob(p)
}) %>%
{grid.arrange(grobs = (.), ncol = 3, nrow = 2)}
or per group (especially important if you used the by argument (interaction in gam).
map(vars, function(x){
p <- plotGAM(gam.college, smooth.cov = x, groupCovs = "Private") +
geom_point(data = train.college, aes_string(y = "Outstate", x = x, color= "Private"), alpha = 0.2) +
geom_rug(data = train.college, aes_string(y = "Outstate", x = x, color= "Private" ), alpha = 0.2) +
scale_color_manual("Private", values = c("#868686FF", "#0073C2FF")) +
theme(legend.position="none")
g <- ggplotGrob(p)
}) %>%
{grid.arrange(grobs = (.), ncol = 3, nrow = 2)}
Update, 08. Jan. 2020.
I currently think the package mgcViz offers superior functionality compared to the voxel::plotGAMfunction. An example using the above data set and models:
library(mgcViz)
viz <- getViz(gam.college)
print(plot(viz, allTerms = T), pages = 1)
plot customization is similar go ggplot2 syntax:
trt <- plot(viz, allTerms = T) +
l_points() +
l_fitLine(linetype = 1) +
l_ciLine(linetype = 3) +
l_ciBar() +
l_rug() +
theme_grey()
print(trt, pages = 1)
This vignette shows many more examples.
Related
So I am getting started with Monte Carlo Sims, and went with this basic code to simulate Returns for a given portfolio. Well somehow a portion of the simulated returns always results in straight linear lines which are easy to see on the plotted graph. First I decreased the number of sims so you can see it clearer and I also played around with some other factors but they keep showing up. The rest of the output looks promising and "random".
Added the link to the image as my account is new and also the code, appreciate any help!:
library(quantmod)
library(ggplot2)
maxDate<- "2000-01-01"
tickers<-c("MSFT", "AAPL", "BRK-B")
getSymbols(tickers, from=maxDate)
Port.p<-na.omit(merge(Cl(AAPL),Cl(MSFT),Cl(`BRK-B`)))
Port.r<-ROC(Port.p, type = "discrete")[-1,]
stock_Price<- as.matrix(Port.p[,1:3])
stock_Returns <- as.matrix(Port.r[,1:3])
mc_rep = 50 # Number of Sims
training_days = 200
portfolio_Weights = c(0.5,0.3,0.2)
coVarMat = cov(stock_Returns)
miu = colMeans(stock_Returns)
Miu = matrix(rep(miu, training_days), nrow = 3)
portfolio_Returns_m = matrix(0, training_days, mc_rep)
set.seed(2000)
for (i in 1:mc_rep) {
Z = matrix ( rnorm( dim(stock_Returns)[2] * training_days ), ncol = training_days )
L = t( chol(coVarMat) )
daily_Returns = Miu + L %*% Z
portfolio_Returns_200 = cumprod( portfolio_Weights %*% daily_Returns + 1 )
portfolio_Returns_m[,i] = portfolio_Returns_200;
}
x_axis = rep(1:training_days, mc_rep)
y_axis = as.vector(portfolio_Returns_m-1)
plot_data = data.frame(x_axis, y_axis)
ggplot(data = plot_data, aes(x = x_axis, y = y_axis)) + geom_path(col = 'red', size = 0.1) +
xlab('Days') + ylab('Portfolio Returns') +
ggtitle('Simulated Portfolio Returns in 200 days')+
theme_bw() +
theme(plot.title = element_text(hjust = 0.5))
The lines are the 'return' from the end of each series to the beginning of the next. You can keep the lines separate by adding a grouping variable to your plotting data and using the group aesthetic to tell ggplot about it:
g <- rep(1:training_days, each = mc_rep)
plot_data = data.frame(x_axis, y_axis, g)
ggplot(data = plot_data, aes(x = x_axis, y = y_axis, group = g)) + ...
Similar questions have been asked before in other forms. Some can be found here and here. However, I cant seem to adapt them when using a facet wrap displaying multiple density plots.
I tried adapting the other examples, but failed... I also tried using the ggpattern package, but when there is a large amount of data, it takes several minutes on my machine to create a plot.
I am trying to create a gradient under the density curve... but with the gradient pointing down. Something like in the example image below:
Some example data to work with:
library(ggplot2)
set.seed(321)
# create data
varNames <- c("x1", "x2", "x3")
df <- data.frame(
var = sample(varNames, 100, replace = T),
val = runif(100)
)
# create plot
ggplot(df, aes(x = val)) +
geom_density(aes(colour = var, fill = var)) +
facet_wrap(~var) +
theme_bw() +
theme(legend.position = "none")
You can use teunbrand's function, but you will need to apply it to each facet. Here simply looping over it with lapply
library(tidyverse)
library(polyclip)
#> polyclip 1.10-0 built from Clipper C++ version 6.4.0
## This is teunbrands function copied without any change!!
## from https://stackoverflow.com/a/64695516/7941188
fade_polygon <- function(x, y, n = 100) {
poly <- data.frame(x = x, y = y)
# Create bounding-box edges
yseq <- seq(min(poly$y), max(poly$y), length.out = n)
xlim <- range(poly$x) + c(-1, 1)
# Pair y-edges
grad <- cbind(head(yseq, -1), tail(yseq, -1))
# Add vertical ID
grad <- cbind(grad, seq_len(nrow(grad)))
# Slice up the polygon
grad <- apply(grad, 1, function(range) {
# Create bounding box
bbox <- data.frame(x = c(xlim, rev(xlim)),
y = c(range[1], range[1:2], range[2]))
# Do actual slicing
slice <- polyclip::polyclip(poly, bbox)
# Format as data.frame
for (i in seq_along(slice)) {
slice[[i]] <- data.frame(
x = slice[[i]]$x,
y = slice[[i]]$y,
value = range[3],
id = c(1, rep(0, length(slice[[i]]$x) - 1))
)
}
slice <- do.call(rbind, slice)
})
# Combine slices
grad <- do.call(rbind, grad)
# Create IDs
grad$id <- cumsum(grad$id)
return(grad)
}
## now here starts the change, loop over your variables. I'm creating the data frame directly instead of keeping the density object
dens <- lapply(split(df, df$var), function(x) {
dens <- density(x$val)
data.frame(x = dens$x, y = dens$y)
}
)
## we need this one for the plot, but still need the list
dens_df <- bind_rows(dens, .id = "var")
grad <- bind_rows(lapply(dens, function(x) fade_polygon(x$x, x$y)), .id = "var")
ggplot(grad, aes(x, y)) +
geom_line(data = dens_df) +
geom_polygon(aes(alpha = value, group = id),
fill = "blue") +
facet_wrap(~var) +
scale_alpha_continuous(range = c(0, 1))
Created on 2021-12-05 by the reprex package (v2.0.1)
We often want individual regression equations in ggplot facets. The best way to do this is build the labels in a dataframe and then add them manually. But what if the labels contain plotmath, e.g., superscripts?
Here is a way to do it. The plotmath is converted to a string and then parsed by ggplot. The test_eqn function is taken from another Stackoverflow post, I'll link it when I find it again. Sorry about that.
library(ggplot2)
library(dplyr)
test_eqn <- function(y, x){
m <- lm(log(y) ~ log(x)) # fit y = a * x ^ b in log space
p <- exp(predict(m)) # model prediction of y
eq <- substitute(expression(Y==a~X^~b),
list(
a = format(unname(exp(coef(m)[1])), digits = 3),
b = format(unname(coef(m)[2]), digits = 3)
))
list(eq = as.character(eq)[2], pred = p)
}
set.seed(123)
x <- runif(20)
y <- runif(20)
test_eqn(x,y)$eq
#> [1] "Y == \"0.57\" ~ X^~\"0.413\""
data <- data.frame(x = x,
y = y,
f = sample(c("A","B"), 20, replace = TRUE)) %>%
group_by(f) %>%
mutate(
label = test_eqn(y,x)$eq, # add label
labelx = mean(x),
labely = mean(y),
pred = test_eqn(y,x)$pred # add prediction
)
# plot fits (use slice(1) to avoid multiple copies of labels)
ggplot(data) +
geom_point(aes(x = x, y = y)) +
geom_line(aes(x = x, y = pred), colour = "red") +
geom_text(data = slice(data, 1), aes(x = labelx, y = labely, label = label), parse = TRUE) +
facet_wrap("f")
Created on 2021-10-20 by the reprex package (v2.0.1)
I would like to plot multiple Poisson (with different lambdas (1:10))
I found the following function to draw a plot
plot_pois = function(lambda = 5)
{
plot(0:20, dpois( x=0:20, lambda=lambda ), xlim=c(-2,20))
normden <- function(x){dnorm(x, mean= lambda, sd=sqrt(lambda))}
curve(normden, from=-4, to=20, add=TRUE, col=lambda)
}
plot.new()
plot_pois(2)
But I can't plot another Poisson over it. I tried to change plot to points or lines but it totally changes the plot. I would also like to add a legends containing different colors for different values of lambda.
If I could plot it using ggplot, it would be a better option.
Another possible tidyverse solution:
library(tidyverse)
# Build Poisson distributions
p_dat <- map_df(1:10, ~ tibble(
l = paste(.),
x = 0:20,
y = dpois(0:20, .)
))
# Build Normal distributions
n_dat <- map_df(1:10, ~ tibble(
l = paste(.),
x = seq(0, 20, by = 0.001),
y = dnorm(seq(0, 20, by = 0.001), ., sqrt(.))
))
# Use ggplot2 to plot
ggplot(n_dat, aes(x, y, color = factor(l, levels = 1:10))) +
geom_line() +
geom_point(data = p_dat, aes(x, y, color = factor(l, levels = 1:10))) +
labs(color = "Lambda:") +
theme_minimal()
Created on 2019-05-06 by the reprex package (v0.2.1)
In ggplot2 you can use lapply to loop over different lambdas:
library(ggplot2)
lambdas <- c(5, 2)
ggplot(data = data.frame(x = 0:20)) +
lapply(lambdas, function(l) geom_point(aes(x = x, y = dpois(x, lambda = l), col = factor(l)))) +
lapply(lambdas, function(l) stat_function(fun = dnorm, args = list(mean = l, sd = sqrt(l)),
aes(x = x, col = factor(l))))
Axes titles and limits, the legend title etc. can then be customized as usual in ggplot2.
I am a beginner at multilevel analysis and try to understand how I can do graphs with the plot functions from base-R. I understand the output of fit below but I am struggeling with the visualization. df is just some simple test data:
t <- seq(0, 10, 1)
df <- data.frame(t = t,
y = 1.5+0.5*(-1)^t + (1.5+0.5*(-1)^t) * t,
p1 = as.factor(rep(c("p1", "p2"), 10)[1:11]))
fit <- lm(y ~ t * p1, data = df)
# I am looking for an automated version of that:
plot(df$t, df$y)
lines(df$t[df$p1 == "p1"],
fit$coefficients[1] + fit$coefficients[2] * df$t[df$p1 == "p1"], col = "blue")
lines(df$t[df$p1 == "p2"],
fit$coefficients[1] + fit$coefficients[2] * df$t[df$p1 == "p2"] +
+ fit$coefficients[3] + fit$coefficients[4] * df$t[df$p1 == "p2"], col = "red")
It should know that it has to include p1 and that there are two lines.
The result should look like this:
Edit: Predict est <- predict(fit, newx = t) gives the same result as fit but still I don't know "how to cluster".
Edit 2 #Keith: The formula y ~ t * p1 reads y = (a + c * p1) + (b + d * p1) * t. For the "first blue line" c, d are both zero.
This is how I would do it. I'm including a ggplot2 version of plot as well because I find it better fitted for the way I think about plots.
This version will account for the number of levels in p1. If you want to compensate for the number of model parameters, you will just have to adjust the way you construct xy to include all the relevant variables. I should point out that if you omit the newdata argument, fitting will be done on the dataset provided to lm.
t <- seq(0, 10, 1)
df <- data.frame(t = t,
y = 1.5+0.5*(-1)^t + (1.5+0.5*(-1)^t) * t,
p1 = as.factor(rep(c("p1", "p2"), 10)[1:11]))
fit <- lm(y ~ t * p1, data = df)
xy <- data.frame(t = t, p1 = rep(levels(df$p1), each = length(t)))
xy$fitted <- predict(fit, newdata = xy)
library(RColorBrewer) # for colors, you can define your own
cols <- brewer.pal(n = length(levels(df$p1)), name = "Set1") # feel free to ignore the warning
plot(x = df$t, y = df$y)
for (i in 1:length(levels(xy$p1))) {
tmp <- xy[xy$p1 == levels(xy$p1)[i], ]
lines(x = tmp$t, y = tmp$fitted, col = cols[i])
}
library(ggplot2)
ggplot(xy, aes(x = t, y = fitted, color = p1)) +
theme_bw() +
geom_point(data = df, aes(x = t, y = y)) +
geom_line()