I am working with the cumulative emergence of flies over time (taken at irregular intervals) over many summers (though first I am just trying to make one year work). The cumulative emergence follows a sigmoid pattern and I want to create a maximum likelihood estimation of a 3-parameter Weibull cumulative distribution function. The three-parameter models I've been trying to use in the fitdistrplus package keep giving me an error. I think this must have something to do with how my data is structured, but I cannot figure it out. Obviously I want it to read each point as an x (degree days) and a y (emergence) value, but it seems to be unable to read two columns. The main error I'm getting says "Non-numeric argument to mathematical function" or (with slightly different code) "data must be a numeric vector of length greater than 1". Below is my code including added columns in the df_dd_em dataframe for cumulative emergence and percent emergence in case that is useful.
degree_days <- c(998.08,1039.66,1111.29,1165.89,1236.53,1293.71,
1347.66,1387.76,1445.47,1493.44,1553.23,1601.97,
1670.28,1737.29,1791.94,1849.20,1920.91,1967.25,
2036.64,2091.85,2152.89,2199.13,2199.13,2263.09,
2297.94,2352.39,2384.03,2442.44,2541.28,2663.90,
2707.36,2773.82,2816.39,2863.94)
emergence <- c(0,0,0,1,1,0,2,3,17,10,0,0,0,2,0,3,0,0,1,5,0,0,0,0,
0,0,0,0,1,0,0,0,0,0)
cum_em <- cumsum(emergence)
df_dd_em <- data.frame (degree_days, emergence, cum_em)
df_dd_em$percent <- ave(df_dd_em$emergence, FUN = function(df_dd_em) 100*(df_dd_em)/46)
df_dd_em$cum_per <- ave(df_dd_em$cum_em, FUN = function(df_dd_em) 100*(df_dd_em)/46)
x <- pweibull(df_dd_em[c(1,3)],shape=5)
dframe2.mle <- fitdist(x, "weibull",method='mle')
Here's my best guess at what you're after:
Set up data:
dd <- data.frame(degree_days=c(998.08,1039.66,1111.29,1165.89,1236.53,1293.71,
1347.66,1387.76,1445.47,1493.44,1553.23,1601.97,
1670.28,1737.29,1791.94,1849.20,1920.91,1967.25,
2036.64,2091.85,2152.89,2199.13,2199.13,2263.09,
2297.94,2352.39,2384.03,2442.44,2541.28,2663.90,
2707.36,2773.82,2816.39,2863.94),
emergence=c(0,0,0,1,1,0,2,3,17,10,0,0,0,2,0,3,0,0,1,5,0,0,0,0,
0,0,0,0,1,0,0,0,0,0))
dd <- transform(dd,cum_em=cumsum(emergence))
We're actually going to fit to an "interval-censored" distribution (i.e. probability of emergence between successive degree day observations: this version assumes that the first observation refers to observations before the first degree-day observation, you could change it to refer to observations after the last observation).
library(bbmle)
## y*log(p) allowing for 0/0 occurrences:
y_log_p <- function(y,p) ifelse(y==0 & p==0,0,y*log(p))
NLLfun <- function(scale,shape,x=dd$degree_days,y=dd$emergence) {
prob <- pmax(diff(pweibull(c(-Inf,x), ## or (c(x,Inf))
shape=shape,scale=scale)),1e-6)
## multinomial probability
-sum(y_log_p(y,prob))
}
library(bbmle)
I should probably have used something more systematic like the method of moments (i.e. matching the mean and variance of a Weibull distribution with the mean and variance of the data), but I just hacked around a bit to find plausible starting values:
## preliminary look (method of moments would be better)
scvec <- 10^(seq(0,4,length=101))
plot(scvec,sapply(scvec,NLLfun,shape=1))
It's important to use parscale to let R know that the parameters are on very different scales:
startvals <- list(scale=1000,shape=1)
m1 <- mle2(NLLfun,start=startvals,
control=list(parscale=unlist(startvals)))
Now try with a three-parameter Weibull (as originally requested) -- requires only a slight modification of what we already have:
library(FAdist)
NLLfun2 <- function(scale,shape,thres,
x=dd$degree_days,y=dd$emergence) {
prob <- pmax(diff(pweibull3(c(-Inf,x),shape=shape,scale=scale,thres)),
1e-6)
## multinomial probability
-sum(y_log_p(y,prob))
}
startvals2 <- list(scale=1000,shape=1,thres=100)
m2 <- mle2(NLLfun2,start=startvals2,
control=list(parscale=unlist(startvals2)))
Looks like the three-parameter fit is much better:
library(emdbook)
AICtab(m1,m2)
## dAIC df
## m2 0.0 3
## m1 21.7 2
And here's the graphical summary:
with(dd,plot(cum_em~degree_days,cex=3))
with(as.list(coef(m1)),curve(sum(dd$emergence)*
pweibull(x,shape=shape,scale=scale),col=2,
add=TRUE))
with(as.list(coef(m2)),curve(sum(dd$emergence)*
pweibull3(x,shape=shape,
scale=scale,thres=thres),col=4,
add=TRUE))
(could also do this more elegantly with ggplot2 ...)
These don't seem like spectacularly good fits, but they're sane. (You could in principle do a chi-squared goodness-of-fit test based on the expected number of emergences per interval, and accounting for the fact that you've fitted a three-parameter model, although the values might be a bit low ...)
Confidence intervals on the fit are a bit of a nuisance; your choices are (1) bootstrapping; (2) parametric bootstrapping (resample parameters assuming a multivariate normal distribution of the data); (3) delta method.
Using bbmle::mle2 makes it easy to do things like get profile confidence intervals:
confint(m1)
## 2.5 % 97.5 %
## scale 1576.685652 1777.437283
## shape 4.223867 6.318481
dd <- data.frame(degree_days=c(998.08,1039.66,1111.29,1165.89,1236.53,1293.71,
1347.66,1387.76,1445.47,1493.44,1553.23,1601.97,
1670.28,1737.29,1791.94,1849.20,1920.91,1967.25,
2036.64,2091.85,2152.89,2199.13,2199.13,2263.09,
2297.94,2352.39,2384.03,2442.44,2541.28,2663.90,
2707.36,2773.82,2816.39,2863.94),
emergence=c(0,0,0,1,1,0,2,3,17,10,0,0,0,2,0,3,0,0,1,5,0,0,0,0,
0,0,0,0,1,0,0,0,0,0))
dd$cum_em <- cumsum(dd$emergence)
dd$percent <- ave(dd$emergence, FUN = function(dd) 100*(dd)/46)
dd$cum_per <- ave(dd$cum_em, FUN = function(dd) 100*(dd)/46)
dd <- transform(dd)
#start 3 parameter model
library(FAdist)
## y*log(p) allowing for 0/0 occurrences:
y_log_p <- function(y,p) ifelse(y==0 & p==0,0,y*log(p))
NLLfun2 <- function(scale,shape,thres,
x=dd$degree_days,y=dd$percent) {
prob <- pmax(diff(pweibull3(c(-Inf,x),shape=shape,scale=scale,thres)),
1e-6)
## multinomial probability
-sum(y_log_p(y,prob))
}
startvals2 <- list(scale=1000,shape=1,thres=100)
m2 <- mle2(NLLfun2,start=startvals2,
control=list(parscale=unlist(startvals2)))
summary(m2)
#graphical summary
windows(5,5)
with(dd,plot(cum_per~degree_days,cex=3))
with(as.list(coef(m2)),curve(sum(dd$percent)*
pweibull3(x,shape=shape,
scale=scale,thres=thres),col=4,
add=TRUE))
Related
I have frequency values changing with the time (x axis units), as presented on the picture below. After some normalization these values may be seen as data points of a density function for some distribution.
Q: Assuming that these frequency points are from Weibull distribution T, how can I fit best Weibull density function to the points so as to infer the distribution T parameters from it?
sample <- c(7787,3056,2359,1759,1819,1189,1077,1080,985,622,648,518,
611,1037,727,489,432,371,1125,69,595,624)
plot(1:length(sample), sample, type = "l")
points(1:length(sample), sample)
Update.
To prevent from being misunderstood, I would like to add little more explanation. By saying I have frequency values changing with the time (x axis units) I mean I have data which says that I have:
7787 realizations of value 1
3056 realizations of value 2
2359 realizations of value 3 ... etc.
Some way towards my goal (incorrect one, as I think) would be to create a set of these realizations:
# Loop to simulate values
set.values <- c()
for(i in 1:length(sample)){
set.values <<- c(set.values, rep(i, times = sample[i]))
}
hist(set.values)
lines(1:length(sample), sample)
points(1:length(sample), sample)
and use fitdistr on the set.values:
f2 <- fitdistr(set.values, 'weibull')
f2
Why I think it is incorrect way and why I am looking for a better solution in R?
in the distribution fitting approach presented above it is assumed that set.values is a complete set of my realisations from the distribution T
in my original question I know the points from the first part of the density curve - I do not know its tail and I want to estimate the tail (and the whole density function)
Here is a better attempt, like before it uses optim to find the best value constrained to a set of values in a box (defined by the lower and upper vectors in the optim call). Notice it scales x and y as part of the optimization in addition to the Weibull distribution shape parameter, so we have 3 parameters to optimize over.
Unfortunately when using all the points it pretty much always finds something on the edges of the constraining box which indicates to me that maybe Weibull is maybe not a good fit for all of the data. The problem is the two points - they ares just too large. You see the attempted fit to all data in the first plot.
If I drop those first two points and just fit the rest, we get a much better fit. You see this in the second plot. I think this is a good fit, it is in any case a local minimum in the interior of the constraining box.
library(optimx)
sample <- c(60953,7787,3056,2359,1759,1819,1189,1077,1080,985,622,648,518,
611,1037,727,489,432,371,1125,69,595,624)
t.sample <- 0:22
s.fit <- sample[3:23]
t.fit <- t.sample[3:23]
wx <- function(param) {
res <- param[2]*dweibull(t.fit*param[3],shape=param[1])
return(res)
}
minwx <- function(param){
v <- s.fit-wx(param)
sqrt(sum(v*v))
}
p0 <- c(1,200,1/20)
paramopt <- optim(p0,minwx,gr=NULL,lower=c(0.1,100,0.01),upper=c(1.1,5000,1))
popt <- paramopt$par
popt
rms <- paramopt$value
tit <- sprintf("Weibull - Shape:%.3f xscale:%.1f yscale:%.5f rms:%.1f",popt[1],popt[2],popt[3],rms)
plot(t.sample[2:23], sample[2:23], type = "p",col="darkred")
lines(t.fit, wx(popt),col="blue")
title(main=tit)
You can directly calculate the maximum likelihood parameters, as described here.
# Defining the error of the implicit function
k.diff <- function(k, vec){
x2 <- seq(length(vec))
abs(k^-1+weighted.mean(log(x2), w = sample)-weighted.mean(log(x2),
w = x2^k*sample))
}
# Setting the error to "quite zero", fulfilling the equation
k <- optimize(k.diff, vec=sample, interval=c(0.1,5), tol=10^-7)$min
# Calculate lambda, given k
l <- weighted.mean(seq(length(sample))^k, w = sample)
# Plot
plot(density(rep(seq(length(sample)),sample)))
x <- 1:25
lines(x, dweibull(x, shape=k, scale= l))
Assuming the data are from a Weibull distribution, you can get an estimate of the shape and scale parameter like this:
sample <- c(7787,3056,2359,1759,1819,1189,1077,1080,985,622,648,518,
611,1037,727,489,432,371,1125,69,595,624)
f<-fitdistr(sample, 'weibull')
f
If you are not sure whether it is distributed Weibull, I would recommend using the ks.test. This tests whether your data is from a hypothesised distribution. Given your knowledge of the nature of the data, you could test for a few selected distributions and see which one works best.
For your example this would look like this:
ks = ks.test(sample, "pweibull", shape=f$estimate[1], scale=f$estimate[2])
ks
The p-value is insignificant, hence you do not reject the hypothesis that the data is from a Weibull distribution.
Update: The histograms of either the Weibull or exponential look like a good match to your data. I think the exponential distribution gives you a better fit. Pareto distribution is another option.
f<-fitdistr(sample, 'weibull')
z<-rweibull(10000, shape= f$estimate[1],scale= f$estimate[2])
hist(z)
f<-fitdistr(sample, 'exponential')
z = rexp(10000, f$estimate[1])
hist(z)
I want to generate 2 continuous random variables Q1, Q2 (quantitative traits, each are normal) and 2 binary random variables Z1, Z2 (binary traits) with given pairwise correlations between all possible pairs of them.
Say
(Q1,Q2):0.23
(Q1,Z1):0.55
(Q1,Z2):0.45
(Q2,Z1):0.4
(Q2,Z2):0.5
(Z1,Z2):0.47
Please help me generate such data in R.
This is crude but might get you started in the right direction.
library(copula)
options(digits=3)
probs <- c(0.5,0.5)
corrs <- c(0.23,0.55,0.45,0.4,0.5,0.47) ## lower triangle
Simulate correlated values (first two quantitative, last two transformed to binary)
sim <- function(n,probs,corrs) {
tmp <- normalCopula( corrs, dim=4 , "un")
getSigma(tmp) ## test
x <- rCopula(1000, tmp)
x2 <- x
x2[,3:4] <- qbinom(x[,3:4],size=1,prob=rep(probs,each=nrow(x)))
x2
}
Test SSQ distance between observed and target correlations:
objfun <- function(corrs,targetcorrs,probs,n=1000) {
cc <- try(cor(sim(n,probs,corrs)),silent=TRUE)
if (is(cc,"try-error")) return(NA)
sum((cc[lower.tri(cc)]-targetcorrs)^2)
}
See how bad things are when input corrs=target:
cc0 <- cor(sim(1000,probs=probs,corrs=corrs))
cc0[lower.tri(cc0)]
corrs
objfun(corrs,corrs,probs=probs) ## 0.112
Now try to optimize.
opt1 <- optim(fn=objfun,
par=corrs,
targetcorrs=corrs,probs=c(0.5,0.5))
opt1$value ## 0.0208
Stops after 501 iterations with "max iterations exceeded". This will never work really well because we're trying to use a deterministic hill-climbing algorithm on a stochastic objective function ...
cc1 <- cor(sim(1000,probs=c(0.5,0.5),corrs=opt1$par))
cc1[lower.tri(cc1)]
corrs
Maybe try simulated annealing?
opt2 <- optim(fn=objfun,
par=corrs,
targetcorrs=corrs,probs=c(0.5,0.5),
method="SANN")
It doesn't seem to do much better than the previous value. Two possible problems (left as an exercise for the reader are) (1) we have specified a set of correlations that are not feasible with the marginal distributions we have chosen, or (2) the error in the objective function surface is getting in the way -- to do better we would have to average over more replicates (i.e. increase n).
I am trying to determine whether there is a significant difference between two Gamm distributions. One distribution has (shape, scale)=(shapeRef,scaleRef) while the other has (shape, scale)=(shapeTarget,scaleTarget). I try to do analysis of variance with the following code
n=10000
x=rgamma(n, shape=shapeRef, scale=scaleRef)
y=rgamma(n, shape=shapeTarget, scale=scaleTarget)
glmm1 <- gam(y~x,family=Gamma(link=log))
anova(glmm1)
The resulting p values keep changing and can be anywhere from <0.1 to >0.9.
Am I going about this the wrong way?
Edit: I use the following code instead
f <- gl(2, n)
x=rgamma(n, shape=shapeRef, scale=scaleRef)
y=rgamma(n, shape=shapeTarget, scale=scaleTarget)
xy <- c(x, y)
anova(glm(xy ~ f, family = Gamma(link = log)),test="F")
But, every time I run it I get a different p-value.
You will indeed get a different p-value every time you run this, if you pick different realizations every time. Just like your data values are random variables, which you'd expect to vary each time you ran an experiment, so is the p-value. If the null hypothesis is true (which was the case in your initial attempts), then the p-values will be uniformly distributed between 0 and 1.
Function to generate simulated data:
simfun <- function(n=100,shapeRef=2,shapeTarget=2,
scaleRef=1,scaleTarget=2) {
f <- gl(2, n)
x=rgamma(n, shape=shapeRef, scale=scaleRef)
y=rgamma(n, shape=shapeTarget, scale=scaleTarget)
xy <- c(x, y)
data.frame(xy,f)
}
Function to run anova() and extract the p-value:
sumfun <- function(d) {
aa <- anova(glm(xy ~ f, family = Gamma(link = log),data=d),test="F")
aa["f","Pr(>F)"]
}
Try it out, 500 times:
set.seed(101)
r <- replicate(500,sumfun(simfun()))
The p-values are always very small (the difference in scale parameters is easily distinguishable), but they do vary:
par(las=1,bty="l") ## cosmetic
hist(log10(r),col="gray",breaks=50)
I have wind data that I'm using to perform extreme value analysis (calculate return levels). I'm using R with packages 'evd', 'extRemes' and 'ismev'.
I'm fitting GEV, Gumbel and Weibull distributions, in order to estimate the return levels (RL) for some period T.
For the GEV and Gumbel cases, I can get RL's and Confidence Intervals using the extRemes::return.level() function.
Some code:
require(ismev)
require(MASS)
data(wind)
x = wind[, 2]
rperiod = 10
fit <- fitdistr(x, 'weibull')
s <- fit$estimate['shape']
b <- fit$estimate['scale']
rlevel <- qweibull(1 - 1/rperiod, shape = s, scale = b)
## CI around rlevel
## ci.rlevel = ??
But for the Weibull case, I need some help to generate the CI's.
I suspect the excruciatingly correct answer will be that the joint confidence region is an ellipse or some bent-sausage shape but you can extract variance estimates for the parameters from the fit object with the vcov function and then build standard errors for which +/- 1.96 SE's should be informative:
> sqrt(vcov(fit)["shape", "shape"])
[1] 0.691422
> sqrt(vcov(fit)["scale", "scale"])
[1] 1.371256
> s +c(-1,1)*sqrt(vcov(fit)["shape", "shape"])
[1] 6.162104 7.544948
> b +c(-1,1)*sqrt(vcov(fit)["scale", "scale"])
[1] 54.46597 57.20848
The usual way to calculate a CI for a single parameter is to assume Normal distribution and use theta+/- 1.96*SE(theta). In this case, you have two parameters so doing that with both of them would give you a "box", the 2D analog of an interval. The truly correct answer would be something more complex in the 'scale'-by-'shape' parameter space and might be most easily achieved with simulation methods, unless you have a better grasp of theory than I have.
In the following code I use bootstrapping to calculate the C.I. and the p-value under the null hypothesis that two different fertilizers applied to tomato plants have no effect in plants yields (and the alternative being that the "improved" fertilizer is better). The first random sample (x) comes from plants where a standard fertilizer has been used, while an "improved" one has been used in the plants where the second sample (y) comes from.
x <- c(11.4,25.3,29.9,16.5,21.1)
y <- c(23.7,26.6,28.5,14.2,17.9,24.3)
total <- c(x,y)
library(boot)
diff <- function(x,i) mean(x[i[6:11]]) - mean(x[i[1:5]])
b <- boot(total, diff, R = 10000)
ci <- boot.ci(b)
p.value <- sum(b$t>=b$t0)/b$R
What I don't like about the code above is that resampling is done as if there was only one sample of 11 values (separating the first 5 as belonging to sample x leaving the rest to sample y).
Could you show me how this code should be modified in order to draw resamples of size 5 with replacement from the first sample and separate resamples of size 6 from the second sample, so that bootstrap resampling would mimic the “separate samples” design that produced the original data?
EDIT2 :
Hack deleted as it was a wrong solution. Instead one has to use the argument strata of the boot function :
total <- c(x,y)
id <- as.factor(c(rep("x",length(x)),rep("y",length(y))))
b <- boot(total, diff, strata=id, R = 10000)
...
Be aware you're not going to get even close to a correct estimate of your p.value :
x <- c(1.4,2.3,2.9,1.5,1.1)
y <- c(23.7,26.6,28.5,14.2,17.9,24.3)
total <- c(x,y)
b <- boot(total, diff, strata=id, R = 10000)
ci <- boot.ci(b)
p.value <- sum(b$t>=b$t0)/b$R
> p.value
[1] 0.5162
How would you explain a p-value of 0.51 for two samples where all values of the second are higher than the highest value of the first?
The above code is fine to get a -biased- estimate of the confidence interval, but the significance testing about the difference should be done by permutation over the complete dataset.
Following John, I think the appropriate way to use bootstrap to test if the sums of these two different populations are significantly different is as follows:
x <- c(1.4,2.3,2.9,1.5,1.1)
y <- c(23.7,26.6,28.5,14.2,17.9,24.3)
b_x <- boot(x, sum, R = 10000)
b_y <- boot(y, sum, R = 10000)
z<-(b_x$t0-b_y$t0)/sqrt(var(b_x$t[,1])+var(b_y$t[,1]))
pnorm(z)
So we can clearly reject the null that they are the same population. I may have missed a degree of freedom adjustment, I am not sure how bootstrapping works in that regard, but such an adjustment will not change your results drastically.
While the actual soil beds could be considered a stratified variable in some instances this is not one of them. You only have the one manipulation, between the groups of plants. Therefore, your null hypothesis is that they really do come from the exact same population. Treating the items as if they're from a single set of 11 samples is the correct way to bootstrap in this case.
If you have two plots, and in each plot tried the different fertilizers over different seasons in a counterbalanced fashion then the plots would be statified samples and you'd want to treat them as such. But that isn't the case here.