scratch/redTime/lookatRedTimeResults_Big1.R
In CollinErickson/CGGP: Composite Grid Gaussian Processes
# This has the results from Big1 and Big2.
# Big1 used UCB and gave bad results.
# Big2 used Greedy and seemed to work well.
# We fixed UCB while it was running, so UCB should work from now on.
x100 <- unname(as.matrix(read.csv("./scratch/redTime/redTimeData/ExpandedRanges2_LHS1L_n100_s0228_all_input.csv")[,-1]))
y100 <- log(unname(as.matrix(read.csv("./scratch/redTime/redTimeData/ExpandedRanges2_LHS1L_n100_s0228_all_output.csv")[,-1])))
x1000 <- unname(as.matrix(read.csv("./scratch/redTime/redTimeData/ExpandedRanges2_LHS1L_n1000_s0303_all_input.csv")[,-1]))
y1000 <- log(unname(as.matrix(read.csv("./scratch/redTime/redTimeData/ExpandedRanges2_LHS1L_n1000_s0303_all_output.csv")[,-1])))
x1000_2 <- unname(as.matrix(read.csv("./scratch/redTime/redTimeData/ExpandedRanges2_LHS1L_n1000_s0304_all_input.csv")[,-1]))
y1000_2 <- log(unname(as.matrix(read.csv("./scratch/redTime/redTimeData/ExpandedRanges2_LHS1L_n1000_s0304_all_output.csv")[,-1])))
# Big1
# Ran SGGP on all outputs, w/ 90 supp pts. Didn't save until 1699 b/c of error
rt.sggp.1699 <- readRDS("./scratch/redTime/redTimeData/out_Big1_SGGP-1699.rds")
rt.sggp.2195 <- readRDS("./scratch/redTime/redTimeData/out_Big1_SGGP-2195.rds")
rt.sggp.2695 <- readRDS("./scratch/redTime/redTimeData/out_Big1_SGGP-2695.rds")
# Big2
r2.sggp.399 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-399.rds")
r2.sggp.599 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-599.rds")
r2.sggp.799 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-799.rds")
r2.sggp.999 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-999.rds")
r2.sggp.1199 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-1199.rds")
r2.sggp.1699 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-1699.rds")
r2.sggp.2199 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-2199.rds")
r2.sggp.2699 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-2699.rds")
r2.sggp.3199 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-3199.rds")
r2.sggp.4199 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-4199.rds")
r2.sggp.5199 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-5199.rds")
r2.sggp.6195 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-6195.rds")
r2.sggp.7187 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-7187.rds")
r2.sggp.8179 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-8179.rds")
r2.sggp.10163<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-10163.rds")
r2.sggp.12159<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-12159.rds")
r2.sggp.16159<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-16159.rds")
r2.sggp.18159<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-18159.rds")
r2.sggp.20155<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-20155.rds")
r2.sggp.22155<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-22155.rds")
r2.sggp.24155<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-24155.rds")
r2.sggp.26155<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-26155.rds")
# r2.sggp.<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-.rds")
r2.sggp.26155.Ignore<- CGGPfit(r2.sggp.26155, Y=r2.sggp.26155$Y,Xs=r2.sggp.26155$Xs,Ys=r2.sggp.26155$Ys,HandlingSuppData="Ignore")
r2.sggp.26155.Mat32<- CGGPfit(r2.sggp.26155, Y=r2.sggp.26155$Y,Xs=r2.sggp.26155$Xs,Ys=r2.sggp.26155$Ys,corr="m3")
r2.sggp.26155.PE<- CGGPfit(r2.sggp.26155, Y=r2.sggp.26155$Y,Xs=r2.sggp.26155$Xs,Ys=r2.sggp.26155$Ys,corr="PowerExp")
r2.sggp.26155.18ktheta <- CGGPfit(r2.sggp.26155, Y=r2.sggp.26155$Y,Xs=r2.sggp.26155$Xs,Ys=r2.sggp.26155$Ys,set_thetaMAP_to = r2.sggp.18159$thetaMAP)
r2.sggp.26155.18kthetanosupp <- CGGPfit(r2.sggp.26155, Y=r2.sggp.26155$Y,set_thetaMAP_to = r2.sggp.18159$thetaMAP)
# Big3
r3.sggp.399 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-399.rds")
r3.sggp.599 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-599.rds")
r3.sggp.799 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-799.rds")
r3.sggp.999 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-999.rds")
r3.sggp.1199 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-1199.rds")
r3.sggp.1699 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-1699.rds")
r3.sggp.2199 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-2199.rds")
r3.sggp.2699 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-2699.rds")
r3.sggp.3199 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-3199.rds")
r3.sggp.4199 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-4199.rds")
r3.sggp.5199 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-5199.rds")
r3.sggp.6195 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-6195.rds")
r3.sggp.7187 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-7187.rds")
r3.sggp.8179 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-8179.rds")
r3.sggp.10163 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-10163.rds")
r3.sggp.12163 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-12163.rds")
r3.sggp.16163 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-16163.rds")
r3.sggp.18163 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-18163.rds")
r3.sggp.20159 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-20159.rds")
r3.sggp.30155 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-30155.rds")
r3.sggp.20159.PE<- CGGPfit(r3.sggp.20159, Y=r3.sggp.20159$Y,Xs=r3.sggp.20159$Xs,Ys=r3.sggp.20159$Ys,corr="PowerExp")
r3.sggp.30155.PE<- CGGPfit(r3.sggp.30155, Y=r3.sggp.30155$Y,Xs=r3.sggp.30155$Xs,Ys=r3.sggp.30155$Ys,corr="PowerExp")
r3.sggp.20159.C <- CGGPfit(r3.sggp.20159, Y=r3.sggp.20159$Y,Xs=r3.sggp.20159$Xs,Ys=r3.sggp.20159$Ys,corr="Cauchy")
r3.sggp.30155.C <- CGGPfit(r3.sggp.30155, Y=r3.sggp.30155$Y,Xs=r3.sggp.30155$Xs,Ys=r3.sggp.30155$Ys,corr="Cauchy")
# Get stats
# Big1
stats.rt.sggp.1699 <- CGGPvalstats(rt.sggp.1699, x1000, y1000, bydim=T)
stats.rt.sggp.2195 <- CGGPvalstats(rt.sggp.2195, x1000, y1000, bydim=T)
stats.rt.sggp.2695 <- CGGPvalstats(rt.sggp.2695, x1000, y1000, bydim=T)
# Big2
stats.r2.sggp.399 <- CGGPvalstats(r2.sggp.399, x1000, y1000, bydim=F)
stats.r2.sggp.599 <- CGGPvalstats(r2.sggp.599, x1000, y1000, bydim=F)
stats.r2.sggp.799 <- CGGPvalstats(r2.sggp.799, x1000, y1000, bydim=F)
stats.r2.sggp.999 <- CGGPvalstats(r2.sggp.999, x1000, y1000, bydim=F)
stats.r2.sggp.1199 <- CGGPvalstats(r2.sggp.1199, x1000, y1000, bydim=F)
stats.r2.sggp.1699 <- CGGPvalstats(r2.sggp.1699, x1000, y1000, bydim=F)
stats.r2.sggp.2199 <- CGGPvalstats(r2.sggp.2199, x1000, y1000, bydim=F)
stats.r2.sggp.2699 <- CGGPvalstats(r2.sggp.2699, x1000, y1000, bydim=F)
stats.r2.sggp.3199 <- CGGPvalstats(r2.sggp.3199, x1000, y1000, bydim=F)
stats.r2.sggp.4199 <- CGGPvalstats(r2.sggp.4199, x1000, y1000, bydim=F)
stats.r2.sggp.5199 <- CGGPvalstats(r2.sggp.5199, x1000, y1000, bydim=F)
stats.r2.sggp.6195 <- CGGPvalstats(r2.sggp.6195, x1000, y1000, bydim=F)
stats.r2.sggp.7187 <- CGGPvalstats(r2.sggp.7187, x1000, y1000, bydim=F)
stats.r2.sggp.8179 <- CGGPvalstats(r2.sggp.8179, x1000, y1000, bydim=F)
stats.r2.sggp.10163<- CGGPvalstats(r2.sggp.10163, x1000, y1000, bydim=F)
stats.r2.sggp.12159<- CGGPvalstats(r2.sggp.12159, x1000, y1000, bydim=F)
stats.r2.sggp.16159<- CGGPvalstats(r2.sggp.18159, x1000, y1000, bydim=F)
stats.r2.sggp.18159<- CGGPvalstats(r2.sggp.18159, x1000, y1000, bydim=F)
stats.r2.sggp.20155<- CGGPvalstats(r2.sggp.20155, x1000, y1000, bydim=F)
stats.r2.sggp.22155<- CGGPvalstats(r2.sggp.22155, x1000, y1000, bydim=F)
stats.r2.sggp.24155<- CGGPvalstats(r2.sggp.24155, x1000, y1000, bydim=F)
stats.r2.sggp.26155<- CGGPvalstats(r2.sggp.26155, x1000, y1000, bydim=F)
# stats.r2.sggp.<- CGGPvalstats(r2.sggp., x1000, y1000, bydim=F)
stats.r2.sggp.26155.Ignore<- CGGPvalstats(r2.sggp.26155.Ignore, x1000, y1000, bydim=F)
stats.r2.sggp.26155.Mat32<- CGGPvalstats(r2.sggp.26155.Mat32, x1000, y1000, bydim=F)
stats.r2.sggp.26155.18ktheta<- CGGPvalstats(r2.sggp.26155.18ktheta, x1000, y1000, bydim=F)
stats.r2.sggp.26155.18kthetanosupp<- CGGPvalstats(r2.sggp.26155.18kthetanosupp, x1000, y1000, bydim=F)
# Big 3
stats.r3.sggp.399 <- CGGPvalstats(r3.sggp.399, x1000, y1000, bydim=F)
stats.r3.sggp.599 <- CGGPvalstats(r3.sggp.599, x1000, y1000, bydim=F)
stats.r3.sggp.799 <- CGGPvalstats(r3.sggp.799, x1000, y1000, bydim=F)
stats.r3.sggp.999 <- CGGPvalstats(r3.sggp.999, x1000, y1000, bydim=F)
stats.r3.sggp.1199 <- CGGPvalstats(r3.sggp.1199, x1000, y1000, bydim=F)
stats.r3.sggp.1699 <- CGGPvalstats(r3.sggp.1699, x1000, y1000, bydim=F)
stats.r3.sggp.2199 <- CGGPvalstats(r3.sggp.2199, x1000, y1000, bydim=F)
stats.r3.sggp.2699 <- CGGPvalstats(r3.sggp.2699, x1000, y1000, bydim=F)
stats.r3.sggp.3199 <- CGGPvalstats(r3.sggp.3199, x1000, y1000, bydim=F)
stats.r3.sggp.4199 <- CGGPvalstats(r3.sggp.4199, x1000, y1000, bydim=F)
stats.r3.sggp.5199 <- CGGPvalstats(r3.sggp.5199, x1000, y1000, bydim=F)
stats.r3.sggp.6195 <- CGGPvalstats(r3.sggp.6195, x1000, y1000, bydim=F)
stats.r3.sggp.7187 <- CGGPvalstats(r3.sggp.7187, x1000, y1000, bydim=F)
stats.r3.sggp.8179 <- CGGPvalstats(r3.sggp.8179, x1000, y1000, bydim=F)
stats.r3.sggp.10163<- CGGPvalstats(r3.sggp.10163, x1000, y1000, bydim=F)
stats.r3.sggp.12163<- CGGPvalstats(r3.sggp.12163, x1000, y1000, bydim=F)
stats.r3.sggp.16163<- CGGPvalstats(r3.sggp.16163, x1000, y1000, bydim=F)
stats.r3.sggp.18163<- CGGPvalstats(r3.sggp.18163, x1000, y1000, bydim=F)
stats.r3.sggp.20159<- CGGPvalstats(r3.sggp.20159, x1000, y1000, bydim=F)
stats.r3.sggp.30155<- CGGPvalstats(r3.sggp.30155, x1000, y1000, bydim=F)
stats.r3.sggp.20159.PE<- CGGPvalstats(r3.sggp.20159.PE, x1000, y1000, bydim=F)
stats.r3.sggp.30155.PE<- CGGPvalstats(r3.sggp.30155.PE, x1000, y1000, bydim=F)
stats.r3.sggp.20159.C <- CGGPvalstats(r3.sggp.20159.C , x1000, y1000, bydim=F)
stats.r3.sggp.30155.C <- CGGPvalstats(r3.sggp.30155.C , x1000, y1000, bydim=F)
# Check stats on 50th dim
CGGPvalstats(CGGPfit(rt.sggp.1699, rt.sggp.1699$Y[,50], Xs=rt.sggp.1699$Xs, Ys=rt.sggp.1699$Ys[,50]), x1000, y1000[,50], bydim=F)
# Run with mlegp
mod.mlegp.50 <- mlegp::mlegp(x100[1:50,], y100[1:50,])
pred.mlegp.50 <- lapply(1:100, function(i) predict(mod.mlegp.50[[i]], x1000, se=T)) %>% {list(fit={do.call(cbind, lapply(., function(i) i$fit))}, se.fit={do.call(cbind, lapply(., function(i) i$se.fit))})}
stats.mlegp.50 <- valstats(pred.mlegp.50$fit, pred.mlegp.50$se, y1000, bydim=F)
mod.mlegp.75 <- mlegp::mlegp(x100[1:75,], y100[1:75,])
pred.mlegp.75 <- lapply(1:100, function(i) predict(mod.mlegp.75[[i]], x1000, se=T)) %>% {list(fit={do.call(cbind, lapply(., function(i) i$fit))}, se.fit={do.call(cbind, lapply(., function(i) i$se.fit))})}
stats.mlegp.75 <- valstats(pred.mlegp.75$fit, pred.mlegp.75$se, y1000, bydim=F)
mod.mlegp.100 <- mlegp::mlegp(x100, y100)
pred.mlegp.100 <- lapply(1:100, function(i) predict(mod.mlegp.100[[i]], x1000, se=T)) %>% {list(fit={do.call(cbind, lapply(., function(i) i$fit))}, se.fit={do.call(cbind, lapply(., function(i) i$se.fit))})}
stats.mlegp.100 <- valstats(pred.mlegp.100$fit, pred.mlegp.100$se, y1000, bydim=F)
mod.mlegp.150 <- sample(1:1000, 150) %>% {mlegp::mlegp(x1000_2[.,], y1000_2[.,])}
pred.mlegp.150 <- lapply(1:100, function(i) predict(mod.mlegp.150[[i]], x1000, se=T)) %>% {list(fit={do.call(cbind, lapply(., function(i) i$fit))}, se.fit={do.call(cbind, lapply(., function(i) i$se.fit))})}
stats.mlegp.150 <- valstats(pred.mlegp.150$fit, pred.mlegp.150$se^2, y1000, bydim=F)
mod.mlegp.200 <- sample(1:1000, 200) %>% {mlegp::mlegp(x1000_2[.,], y1000_2[.,])}
pred.mlegp.200 <- lapply(1:100, function(i) predict(mod.mlegp.200[[i]], x1000, se=T)) %>% {list(fit={do.call(cbind, lapply(., function(i) i$fit))}, se.fit={do.call(cbind, lapply(., function(i) i$se.fit))})}
stats.mlegp.200 <- valstats(pred.mlegp.200$fit, pred.mlegp.200$se^2, y1000, bydim=F)
mod.mlegp.250 <- sample(1:1000, 250) %>% {mlegp::mlegp(x1000_2[.,], y1000_2[.,])}
pred.mlegp.250 <- lapply(1:100, function(i) predict(mod.mlegp.250[[i]], x1000, se=T)) %>% {list(fit={do.call(cbind, lapply(., function(i) i$fit))}, se.fit={do.call(cbind, lapply(., function(i) i$se.fit))})}
stats.mlegp.250 <- valstats(pred.mlegp.250$fit, pred.mlegp.250$se^2, y1000, bydim=F)
# mod.mlegp.300 <- mlegp::mlegp(x1000_2, y1000_2)
# pred.mlegp.300 <- predict(mod.mlegp.300, x1000, se=T)
# stats.mlegp.300 <- valstats(pred.mlegp.300$fit, pred.mlegp.300$se^2, y1000, bydim=F)
# mod.mlegp.400 <- mlegp::mlegp(x1000_2, y1000_2)
# pred.mlegp.400 <- predict(mod.mlegp.400, x1000, se=T)
# stats.mlegp.400 <- valstats(pred.mlegp.400$fit, pred.mlegp.400$se^2, y1000, bydim=F)
# mod.mlegp.500 <- mlegp::mlegp(x1000_2, y1000_2)
# pred.mlegp.500 <- predict(mod.mlegp.500, x1000, se=T)
# stats.mlegp.500 <- valstats(pred.mlegp.500$fit, pred.mlegp.500$se^2, y1000, bydim=F)
allstats <- list(
# CGGP
data.frame("CGGP1", 90, 1699, 0.03876258, -5.951558, 0.01723906, 0.92504, 0.9998292, 0.9996583, 0.02941151),
data.frame("CGGP1", 90, 2195, 0.03973626, -5.939358, 0.01805724, 0.94663, 0.9998205, 0.9996409, 0.0301442),
data.frame("CGGP1", 90, 2695, 0.03895927, -5.709819, 0.01752450, 0.89132, 0.9998274, 0.9996548, 0.02955428),
data.frame("CGGP", 90, 399, 0.03292523, -6.292347, 0.01476177, 0.93929, 0.9998769, 0.9997534, 0.02502954),
data.frame("CGGP", 90, 599, 0.02445636, -6.717207, 0.01163083, 0.95573, 0.999932, 0.999864, 0.01859961),
data.frame("CGGP", 90, 799, 0.02426409, -6.831546, 0.01119061, 0.95034, 0.9999334, 0.9998661, 0.01841507),
data.frame("CGGP", 90, 999, 0.02418435, -6.846158, 0.01062984, 0.95242, 0.9999338, 0.999867, 0.01836773),
data.frame("CGGP", 90, 1199, 0.02205223, -7.085375, 0.009788555, 0.95352, 0.9999454, 0.9998894, 0.01662363),
data.frame("CGGP", 90, 1699, 0.01527591, -7.696946, 0.007269341, 0.97937, 0.9999735, 0.9999469, 0.01145086),
data.frame("CGGP", 90, 2199, 0.01370636, -7.730417, 0.006900497, 0.9812, 0.9999787, 0.9999573, 0.01036107),
data.frame("CGGP", 90, 2699, 0.014988 , -7.711674, 0.006816212, 0.98524, 0.9999745, 0.9999489, 0.01092403),
data.frame("CGGP", 90, 3199, 0.01357989, -8.056265, 0.005935432, 0.98193, 0.9999791, 0.9999581, 0.009891518),
data.frame("CGGP", 90, 4199, 0.01130804, -8.147965, 0.005525203, 0.97767, 0.9999855, 0.9999709, 0.008477671),
data.frame("CGGP", 90, 5199, 0.009739982,-8.454266, 0.004678459, 0.98001, 0.9999892, 0.9999784, 0.007307837),
data.frame("CGGP", 90, 6195, 0.007799561, -8.601915, 0.004069435, 0.99209, 0.9999931, 0.9999862, 0.005720459),
data.frame("CGGP", 90, 7187, 0.007602274, -8.832714, 0.003677643, 0.9891, 0.9999934, 0.9999869, 0.005552559),
data.frame("CGGP", 90, 8179, 0.006772034, -9.045547, 0.003310222, 0.98978, 0.9999948, 0.9999896, 0.004955794),
data.frame("CGGP", 90, 10163,0.005923366, -9.315209, 0.002928836, 0.98941, 0.999996, 0.999992, 0.004356485),
data.frame("CGGP", 90, 12159,0.004279414, -9.339384, 0.002790037, 0.99488, 0.9999979, 0.9999958, 0.003197408),
data.frame("CGGP", 90, 16159,0.003443654, -9.731616, 0.002267292, 0.99656, 0.9999987, 0.9999973, 0.002545515),
data.frame("CGGP", 90, 18159,0.003443654, -9.731616, 0.002267292, 0.99656, 0.9999987, 0.9999973, 0.002545515),
data.frame("CGGP", 90, 20155,0.003984703, -9.738379, 0.002263422, 0.99591, 0.9999982, 0.9999964, 0.002905657),
data.frame("CGGP", 90, 22155,0.004459119, -9.739725, 0.002274312, 0.99623, 0.9999977, 0.9999955, 0.00319555),
data.frame("CGGP", 90, 24155,0.004439284, -9.703938, 0.002317367, 0.99618, 0.9999978, 0.9999955, 0.003193208),
data.frame("CGGP", 90, 26155,0.004961369, -9.659113, 0.002383479, 0.99554, 0.9999972, 0.9999944, 0.00355455),
data.frame("CGGPIgnore", 90, 26155,0.004959593, -9.656147, 0.002386668, 0.99554, 0.9999972, 0.9999944, 0.003553383),
data.frame("CGGPMat32", 90, 26155, 0.004838821, -8.49521 , 0.003830926, 0.99942, 0.9999973, 0.9999947, 0.003493466),
data.frame("CGGP18ktheta", 90, 26155, 0.004768937, -7.572385, 0.005782846, 0.99923, 0.9999974, 0.9999948, 0.003410945),
data.frame("CGGP18kthetanosupp", 90, 26155, 0.00473138, -7.165235, 0.007330487, 0.99932, 0.9999975, 0.9999949, 0.003383729),
# data.frame("CGGP", 90, 2199, ),
# data.frame("CGGP", 90, , ),
# data.frame("CGGP", 90, , ),
data.frame("CGGP3", 90, 399, 0.03086891, -6.507128, 0.01359139, 0.9464, 0.9998919, 0.9997833, 0.023462),
data.frame("CGGP3", 90, 599, 0.02465403, -6.811887, 0.01158544, 0.94892, 0.999931, 0.9998618, 0.01870552),
data.frame("CGGP3", 90, 799, 0.02501423, -6.701828, 0.01198723, 0.9419, 0.9999293, 0.9998577, 0.01896552),
data.frame("CGGP3", 90, 999, 0.02635975, -6.529451, 0.01207771, 0.90545, 0.9999212, 0.999842 , 0.02004152),
data.frame("CGGP3", 90, 1199, 0.02474022, -6.754248, 0.01105155, 0.90884, 0.9999306, 0.9998608, 0.01878582),
data.frame("CGGP3", 90, 1699, 0.01494981, -7.536539, 0.007537477, 0.97898, 0.9999747, 0.9999492, 0.01117594),
data.frame("CGGP3", 90, 2199, 0.01467389, -7.641863, 0.00727886, 0.97162, 0.9999755, 0.999951 , 0.01102318),
data.frame("CGGP3", 90, 2699, 0.01491939, -7.712265, 0.006820505, 0.98566, 0.9999747, 0.9999494, 0.01086603),
data.frame("CGGP3", 90, 3199, 0.01306396, -8.091254, 0.005814433, 0.98243, 0.9999806, 0.9999612, 0.009469989),
data.frame("CGGP3", 90, 4199, 0.01082473, -8.154173, 0.005511459, 0.98191, 0.9999867, 0.9999734, 0.008157038),
data.frame("CGGP3", 90, 5199, 0.00822481, -8.572066, 0.004325123, 0.98704, 0.9999923, 0.9999846, 0.006089218),
data.frame("CGGP3", 90, 6195, 0.008069555, -8.623623, 0.004043901, 0.99013, 0.9999926, 0.9999852, 0.005929349),
data.frame("CGGP3", 90, 7187, 0.007608375, -8.91085 , 0.003599635, 0.98793, 0.9999934, 0.9999868, 0.005580803),
data.frame("CGGP3", 90, 8179, 0.00736447, -9.108104, 0.003316512, 0.98517, 0.9999938, 0.9999877, 0.005399971),
data.frame("CGGP3", 90, 10163, 0.006123244, -9.453641, 0.002815331, 0.98794, 0.9999958, 0.9999915, 0.004529484),
data.frame("CGGP3", 90, 12163, 0.004061737, -9.652619, 0.002411071, 0.99395, 0.9999981, 0.9999962, 0.003012229),
data.frame("CGGP3", 90, 16163, 0.004127308, -9.924129, 0.002172679, 0.99046, 0.9999981, 0.9999961, 0.003106536),
data.frame("CGGP3", 90, 18163, 0.003674585, -10.05135, 0.002041532, 0.99177, 0.9999985, 0.9999969, 0.002787224),
data.frame("CGGP3", 90, 20159, 0.003491477, -10.04612, 0.001995089, 0.99532, 0.9999986, 0.9999972, 0.002585274),
data.frame("CGGP3", 90, 30155, 0.004722018, -10.10524, 0.00193077, 0.99323, 0.9999975, 0.9999949, 0.003389804),
data.frame("CGGP3PE",90,20159, 0.003089233, -8.211388, 0.004281498, 1, 0.9999989, 0.9999978, 0.002299494),
data.frame("CGGP3PE",90,30155, 0.003431799, -8.39904, 0.003919711, 0.99961, 0.9999987, 0.9999973, 0.002488821),
data.frame("CGGP3C" ,90,20159, 0.003050064, -8.897864, 0.003109479, 0.99982, 0.9999989, 0.9999979, 0.002269927),
data.frame("CGGP3C" ,90,30155, 0.003797589, -8.963827, 0.003034772, 0.99897, 0.9999984, 0.9999967, 0.002741234),
# data.frame("CGGP3", 90, , ),
# mlegp
data.frame("mlegp", 0, 50, 0.2121086, -2.136166, 0.1122491, 0.9941, 0.9949418, 0.9897676, 0.1614444),
data.frame("mlegp", 0, 75, 0.1266577, -2.708455, 0.0773198, 0.99936, 0.9982067, 0.9963514, 0.09525388),
data.frame("mlegp", 0,100, 0.1005995, -2.896368, 0.06750803, 0.9998, 0.9988543, 0.9976983, 0.07583394),
data.frame("mlegp", 0,150, 0.08580729,-4.263628, 0.04071472, 0.83153, 0.9991757, 0.9983254, 0.06421618),
data.frame("mlegp", 0,200, 0.06383321, -4.68457, 0.02916151, 0.79007, 0.9995385, 0.9990733, 0.04804422),
data.frame("mlegp", 0,250, 0.05941459,-5.003207, 0.02610797, 0.80917, 0.9996011, 0.9991971, 0.04498041)
)
allstats <- lapply(allstats, function(x){colnames(x) <- c("Package", 'Nsup',"Ngrid","RMSE","score","CRPscore","coverage","corr","R2","RMSEnorm");x})
allstats <- do.call(rbind, allstats)
allstats$Ntotal <- allstats$Nsup + allstats$Ngrid
allstats$Package <- as.character(allstats$Package)
library(ggplot2)
# ggplot(data=allstats, mapping=aes(Ntotal, RMSE, color=Package, shape=as.factor(Nsup))) + geom_point(size=3) + scale_y_log10()
ggplot(data=allstats, mapping=aes(Ntotal, RMSE, color=Package, shape=as.factor(Nsup))) + geom_point(size=3) + scale_x_log10() + scale_y_log10()
ggplot(data=allstats, mapping=aes(Ntotal, RMSE, color=Nsup)) + geom_point(size=3) + facet_grid(. ~ Package) + scale_x_log10() + scale_y_log10()
ggplot(data=allstats, mapping=aes(Ntotal, score, color=(Package), shape=as.factor(Nsup))) + geom_point(size=3) + scale_x_log10()
ggplot(data=allstats, mapping=aes(Ntotal, CRPscore, color=(Package), shape=as.factor(Nsup))) + geom_point(size=3) + scale_x_log10() + scale_y_log10()
# ggplot(data=allstats %>% filter(score<1e5), mapping=aes(Ntotal, score, color=(Package), shape=as.factor(Nsup))) + geom_point(size=3) + scale_x_log10()
tdf <- rbind(cbind(Ngrid=1699, stats.rt.sggp.1699),
cbind(Ngrid=2195, stats.rt.sggp.2195),
cbind(Ngrid=2695, stats.rt.sggp.2695))
ggplot(data=tdf, mapping=aes(Ngrid, RMSE)) + geom_point()
# Plot for thesis
tdf2 <- allstats %>% filter(Package %in% c("CGGP3", "mlegp"))
tdf2$Package[tdf2$Package == "CGGP3"] <- "CGGP"
ggrmse <- ggplot(data=tdf2, mapping=aes(Ntotal, RMSE, color=Package, shape=Package)) + geom_point(size=4) + scale_x_log10() + scale_y_log10() + xlab("Number of points evaluated"); ggrmse
maybe_save("redTimeRMSE", ggrmse, height=6, width=7)
# Plots for dissertation for specific redTime object
r3.sggp.20159 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-20159.rds")
# blocks plot
r3.sggp.20159 %>% CGGPplotblocks() + coord_fixed()
maybe_save("redTime_plotblocks", r3.sggp.20159 %>% CGGPplotblocks() + coord_fixed(), height=6, width=6)
# Correlation plot
r3.sggp.20159 %>% CGGPplotcorr()
maybe_save("redTime_corr", r3.sggp.20159 %>% CGGPplotcorr(), height=8, width=6.8)
# Projection plot
r3.sggp.20159 %>% CGGPplotprojection(outdims = c(20,50,80))
r3.sggp.20159 %>% CGGPplotprojection(outdims = c(50))
r3.sggp.20159 %>% CGGPplotprojection(outdims = c(50), facet="wrap")
maybe_save("redTimeProjection", r3.sggp.20159 %>% CGGPplotprojection(outdims = c(50), facet="wrap"), height=6, width=6)
CollinErickson/CGGP documentation built on Feb. 6, 2024, 2:24 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
# This has the results from Big1 and Big2.
# Big1 used UCB and gave bad results.
# Big2 used Greedy and seemed to work well.
# We fixed UCB while it was running, so UCB should work from now on.
x100 <- unname(as.matrix(read.csv("./scratch/redTime/redTimeData/ExpandedRanges2_LHS1L_n100_s0228_all_input.csv")[,-1]))
y100 <- log(unname(as.matrix(read.csv("./scratch/redTime/redTimeData/ExpandedRanges2_LHS1L_n100_s0228_all_output.csv")[,-1])))
x1000 <- unname(as.matrix(read.csv("./scratch/redTime/redTimeData/ExpandedRanges2_LHS1L_n1000_s0303_all_input.csv")[,-1]))
y1000 <- log(unname(as.matrix(read.csv("./scratch/redTime/redTimeData/ExpandedRanges2_LHS1L_n1000_s0303_all_output.csv")[,-1])))
x1000_2 <- unname(as.matrix(read.csv("./scratch/redTime/redTimeData/ExpandedRanges2_LHS1L_n1000_s0304_all_input.csv")[,-1]))
y1000_2 <- log(unname(as.matrix(read.csv("./scratch/redTime/redTimeData/ExpandedRanges2_LHS1L_n1000_s0304_all_output.csv")[,-1])))
# Big1
# Ran SGGP on all outputs, w/ 90 supp pts. Didn't save until 1699 b/c of error
rt.sggp.1699 <- readRDS("./scratch/redTime/redTimeData/out_Big1_SGGP-1699.rds")
rt.sggp.2195 <- readRDS("./scratch/redTime/redTimeData/out_Big1_SGGP-2195.rds")
rt.sggp.2695 <- readRDS("./scratch/redTime/redTimeData/out_Big1_SGGP-2695.rds")
# Big2
r2.sggp.399 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-399.rds")
r2.sggp.599 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-599.rds")
r2.sggp.799 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-799.rds")
r2.sggp.999 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-999.rds")
r2.sggp.1199 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-1199.rds")
r2.sggp.1699 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-1699.rds")
r2.sggp.2199 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-2199.rds")
r2.sggp.2699 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-2699.rds")
r2.sggp.3199 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-3199.rds")
r2.sggp.4199 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-4199.rds")
r2.sggp.5199 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-5199.rds")
r2.sggp.6195 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-6195.rds")
r2.sggp.7187 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-7187.rds")
r2.sggp.8179 <- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-8179.rds")
r2.sggp.10163<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-10163.rds")
r2.sggp.12159<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-12159.rds")
r2.sggp.16159<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-16159.rds")
r2.sggp.18159<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-18159.rds")
r2.sggp.20155<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-20155.rds")
r2.sggp.22155<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-22155.rds")
r2.sggp.24155<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-24155.rds")
r2.sggp.26155<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-26155.rds")
# r2.sggp.<- readRDS("./scratch/redTime/redTimeData/out_Big2_SGGP-.rds")
r2.sggp.26155.Ignore<- CGGPfit(r2.sggp.26155, Y=r2.sggp.26155$Y,Xs=r2.sggp.26155$Xs,Ys=r2.sggp.26155$Ys,HandlingSuppData="Ignore")
r2.sggp.26155.Mat32<- CGGPfit(r2.sggp.26155, Y=r2.sggp.26155$Y,Xs=r2.sggp.26155$Xs,Ys=r2.sggp.26155$Ys,corr="m3")
r2.sggp.26155.PE<- CGGPfit(r2.sggp.26155, Y=r2.sggp.26155$Y,Xs=r2.sggp.26155$Xs,Ys=r2.sggp.26155$Ys,corr="PowerExp")
r2.sggp.26155.18ktheta <- CGGPfit(r2.sggp.26155, Y=r2.sggp.26155$Y,Xs=r2.sggp.26155$Xs,Ys=r2.sggp.26155$Ys,set_thetaMAP_to = r2.sggp.18159$thetaMAP)
r2.sggp.26155.18kthetanosupp <- CGGPfit(r2.sggp.26155, Y=r2.sggp.26155$Y,set_thetaMAP_to = r2.sggp.18159$thetaMAP)
# Big3
r3.sggp.399 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-399.rds")
r3.sggp.599 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-599.rds")
r3.sggp.799 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-799.rds")
r3.sggp.999 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-999.rds")
r3.sggp.1199 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-1199.rds")
r3.sggp.1699 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-1699.rds")
r3.sggp.2199 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-2199.rds")
r3.sggp.2699 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-2699.rds")
r3.sggp.3199 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-3199.rds")
r3.sggp.4199 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-4199.rds")
r3.sggp.5199 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-5199.rds")
r3.sggp.6195 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-6195.rds")
r3.sggp.7187 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-7187.rds")
r3.sggp.8179 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-8179.rds")
r3.sggp.10163 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-10163.rds")
r3.sggp.12163 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-12163.rds")
r3.sggp.16163 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-16163.rds")
r3.sggp.18163 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-18163.rds")
r3.sggp.20159 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-20159.rds")
r3.sggp.30155 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-30155.rds")
r3.sggp.20159.PE<- CGGPfit(r3.sggp.20159, Y=r3.sggp.20159$Y,Xs=r3.sggp.20159$Xs,Ys=r3.sggp.20159$Ys,corr="PowerExp")
r3.sggp.30155.PE<- CGGPfit(r3.sggp.30155, Y=r3.sggp.30155$Y,Xs=r3.sggp.30155$Xs,Ys=r3.sggp.30155$Ys,corr="PowerExp")
r3.sggp.20159.C <- CGGPfit(r3.sggp.20159, Y=r3.sggp.20159$Y,Xs=r3.sggp.20159$Xs,Ys=r3.sggp.20159$Ys,corr="Cauchy")
r3.sggp.30155.C <- CGGPfit(r3.sggp.30155, Y=r3.sggp.30155$Y,Xs=r3.sggp.30155$Xs,Ys=r3.sggp.30155$Ys,corr="Cauchy")
# Get stats
# Big1
stats.rt.sggp.1699 <- CGGPvalstats(rt.sggp.1699, x1000, y1000, bydim=T)
stats.rt.sggp.2195 <- CGGPvalstats(rt.sggp.2195, x1000, y1000, bydim=T)
stats.rt.sggp.2695 <- CGGPvalstats(rt.sggp.2695, x1000, y1000, bydim=T)
# Big2
stats.r2.sggp.399 <- CGGPvalstats(r2.sggp.399, x1000, y1000, bydim=F)
stats.r2.sggp.599 <- CGGPvalstats(r2.sggp.599, x1000, y1000, bydim=F)
stats.r2.sggp.799 <- CGGPvalstats(r2.sggp.799, x1000, y1000, bydim=F)
stats.r2.sggp.999 <- CGGPvalstats(r2.sggp.999, x1000, y1000, bydim=F)
stats.r2.sggp.1199 <- CGGPvalstats(r2.sggp.1199, x1000, y1000, bydim=F)
stats.r2.sggp.1699 <- CGGPvalstats(r2.sggp.1699, x1000, y1000, bydim=F)
stats.r2.sggp.2199 <- CGGPvalstats(r2.sggp.2199, x1000, y1000, bydim=F)
stats.r2.sggp.2699 <- CGGPvalstats(r2.sggp.2699, x1000, y1000, bydim=F)
stats.r2.sggp.3199 <- CGGPvalstats(r2.sggp.3199, x1000, y1000, bydim=F)
stats.r2.sggp.4199 <- CGGPvalstats(r2.sggp.4199, x1000, y1000, bydim=F)
stats.r2.sggp.5199 <- CGGPvalstats(r2.sggp.5199, x1000, y1000, bydim=F)
stats.r2.sggp.6195 <- CGGPvalstats(r2.sggp.6195, x1000, y1000, bydim=F)
stats.r2.sggp.7187 <- CGGPvalstats(r2.sggp.7187, x1000, y1000, bydim=F)
stats.r2.sggp.8179 <- CGGPvalstats(r2.sggp.8179, x1000, y1000, bydim=F)
stats.r2.sggp.10163<- CGGPvalstats(r2.sggp.10163, x1000, y1000, bydim=F)
stats.r2.sggp.12159<- CGGPvalstats(r2.sggp.12159, x1000, y1000, bydim=F)
stats.r2.sggp.16159<- CGGPvalstats(r2.sggp.18159, x1000, y1000, bydim=F)
stats.r2.sggp.18159<- CGGPvalstats(r2.sggp.18159, x1000, y1000, bydim=F)
stats.r2.sggp.20155<- CGGPvalstats(r2.sggp.20155, x1000, y1000, bydim=F)
stats.r2.sggp.22155<- CGGPvalstats(r2.sggp.22155, x1000, y1000, bydim=F)
stats.r2.sggp.24155<- CGGPvalstats(r2.sggp.24155, x1000, y1000, bydim=F)
stats.r2.sggp.26155<- CGGPvalstats(r2.sggp.26155, x1000, y1000, bydim=F)
# stats.r2.sggp.<- CGGPvalstats(r2.sggp., x1000, y1000, bydim=F)
stats.r2.sggp.26155.Ignore<- CGGPvalstats(r2.sggp.26155.Ignore, x1000, y1000, bydim=F)
stats.r2.sggp.26155.Mat32<- CGGPvalstats(r2.sggp.26155.Mat32, x1000, y1000, bydim=F)
stats.r2.sggp.26155.18ktheta<- CGGPvalstats(r2.sggp.26155.18ktheta, x1000, y1000, bydim=F)
stats.r2.sggp.26155.18kthetanosupp<- CGGPvalstats(r2.sggp.26155.18kthetanosupp, x1000, y1000, bydim=F)
# Big 3
stats.r3.sggp.399 <- CGGPvalstats(r3.sggp.399, x1000, y1000, bydim=F)
stats.r3.sggp.599 <- CGGPvalstats(r3.sggp.599, x1000, y1000, bydim=F)
stats.r3.sggp.799 <- CGGPvalstats(r3.sggp.799, x1000, y1000, bydim=F)
stats.r3.sggp.999 <- CGGPvalstats(r3.sggp.999, x1000, y1000, bydim=F)
stats.r3.sggp.1199 <- CGGPvalstats(r3.sggp.1199, x1000, y1000, bydim=F)
stats.r3.sggp.1699 <- CGGPvalstats(r3.sggp.1699, x1000, y1000, bydim=F)
stats.r3.sggp.2199 <- CGGPvalstats(r3.sggp.2199, x1000, y1000, bydim=F)
stats.r3.sggp.2699 <- CGGPvalstats(r3.sggp.2699, x1000, y1000, bydim=F)
stats.r3.sggp.3199 <- CGGPvalstats(r3.sggp.3199, x1000, y1000, bydim=F)
stats.r3.sggp.4199 <- CGGPvalstats(r3.sggp.4199, x1000, y1000, bydim=F)
stats.r3.sggp.5199 <- CGGPvalstats(r3.sggp.5199, x1000, y1000, bydim=F)
stats.r3.sggp.6195 <- CGGPvalstats(r3.sggp.6195, x1000, y1000, bydim=F)
stats.r3.sggp.7187 <- CGGPvalstats(r3.sggp.7187, x1000, y1000, bydim=F)
stats.r3.sggp.8179 <- CGGPvalstats(r3.sggp.8179, x1000, y1000, bydim=F)
stats.r3.sggp.10163<- CGGPvalstats(r3.sggp.10163, x1000, y1000, bydim=F)
stats.r3.sggp.12163<- CGGPvalstats(r3.sggp.12163, x1000, y1000, bydim=F)
stats.r3.sggp.16163<- CGGPvalstats(r3.sggp.16163, x1000, y1000, bydim=F)
stats.r3.sggp.18163<- CGGPvalstats(r3.sggp.18163, x1000, y1000, bydim=F)
stats.r3.sggp.20159<- CGGPvalstats(r3.sggp.20159, x1000, y1000, bydim=F)
stats.r3.sggp.30155<- CGGPvalstats(r3.sggp.30155, x1000, y1000, bydim=F)
stats.r3.sggp.20159.PE<- CGGPvalstats(r3.sggp.20159.PE, x1000, y1000, bydim=F)
stats.r3.sggp.30155.PE<- CGGPvalstats(r3.sggp.30155.PE, x1000, y1000, bydim=F)
stats.r3.sggp.20159.C <- CGGPvalstats(r3.sggp.20159.C , x1000, y1000, bydim=F)
stats.r3.sggp.30155.C <- CGGPvalstats(r3.sggp.30155.C , x1000, y1000, bydim=F)
# Check stats on 50th dim
CGGPvalstats(CGGPfit(rt.sggp.1699, rt.sggp.1699$Y[,50], Xs=rt.sggp.1699$Xs, Ys=rt.sggp.1699$Ys[,50]), x1000, y1000[,50], bydim=F)
# Run with mlegp
mod.mlegp.50 <- mlegp::mlegp(x100[1:50,], y100[1:50,])
pred.mlegp.50 <- lapply(1:100, function(i) predict(mod.mlegp.50[[i]], x1000, se=T)) %>% {list(fit={do.call(cbind, lapply(., function(i) i$fit))}, se.fit={do.call(cbind, lapply(., function(i) i$se.fit))})}
stats.mlegp.50 <- valstats(pred.mlegp.50$fit, pred.mlegp.50$se, y1000, bydim=F)
mod.mlegp.75 <- mlegp::mlegp(x100[1:75,], y100[1:75,])
pred.mlegp.75 <- lapply(1:100, function(i) predict(mod.mlegp.75[[i]], x1000, se=T)) %>% {list(fit={do.call(cbind, lapply(., function(i) i$fit))}, se.fit={do.call(cbind, lapply(., function(i) i$se.fit))})}
stats.mlegp.75 <- valstats(pred.mlegp.75$fit, pred.mlegp.75$se, y1000, bydim=F)
mod.mlegp.100 <- mlegp::mlegp(x100, y100)
pred.mlegp.100 <- lapply(1:100, function(i) predict(mod.mlegp.100[[i]], x1000, se=T)) %>% {list(fit={do.call(cbind, lapply(., function(i) i$fit))}, se.fit={do.call(cbind, lapply(., function(i) i$se.fit))})}
stats.mlegp.100 <- valstats(pred.mlegp.100$fit, pred.mlegp.100$se, y1000, bydim=F)
mod.mlegp.150 <- sample(1:1000, 150) %>% {mlegp::mlegp(x1000_2[.,], y1000_2[.,])}
pred.mlegp.150 <- lapply(1:100, function(i) predict(mod.mlegp.150[[i]], x1000, se=T)) %>% {list(fit={do.call(cbind, lapply(., function(i) i$fit))}, se.fit={do.call(cbind, lapply(., function(i) i$se.fit))})}
stats.mlegp.150 <- valstats(pred.mlegp.150$fit, pred.mlegp.150$se^2, y1000, bydim=F)
mod.mlegp.200 <- sample(1:1000, 200) %>% {mlegp::mlegp(x1000_2[.,], y1000_2[.,])}
pred.mlegp.200 <- lapply(1:100, function(i) predict(mod.mlegp.200[[i]], x1000, se=T)) %>% {list(fit={do.call(cbind, lapply(., function(i) i$fit))}, se.fit={do.call(cbind, lapply(., function(i) i$se.fit))})}
stats.mlegp.200 <- valstats(pred.mlegp.200$fit, pred.mlegp.200$se^2, y1000, bydim=F)
mod.mlegp.250 <- sample(1:1000, 250) %>% {mlegp::mlegp(x1000_2[.,], y1000_2[.,])}
pred.mlegp.250 <- lapply(1:100, function(i) predict(mod.mlegp.250[[i]], x1000, se=T)) %>% {list(fit={do.call(cbind, lapply(., function(i) i$fit))}, se.fit={do.call(cbind, lapply(., function(i) i$se.fit))})}
stats.mlegp.250 <- valstats(pred.mlegp.250$fit, pred.mlegp.250$se^2, y1000, bydim=F)
# mod.mlegp.300 <- mlegp::mlegp(x1000_2, y1000_2)
# pred.mlegp.300 <- predict(mod.mlegp.300, x1000, se=T)
# stats.mlegp.300 <- valstats(pred.mlegp.300$fit, pred.mlegp.300$se^2, y1000, bydim=F)
# mod.mlegp.400 <- mlegp::mlegp(x1000_2, y1000_2)
# pred.mlegp.400 <- predict(mod.mlegp.400, x1000, se=T)
# stats.mlegp.400 <- valstats(pred.mlegp.400$fit, pred.mlegp.400$se^2, y1000, bydim=F)
# mod.mlegp.500 <- mlegp::mlegp(x1000_2, y1000_2)
# pred.mlegp.500 <- predict(mod.mlegp.500, x1000, se=T)
# stats.mlegp.500 <- valstats(pred.mlegp.500$fit, pred.mlegp.500$se^2, y1000, bydim=F)
allstats <- list(
# CGGP
data.frame("CGGP1", 90, 1699, 0.03876258, -5.951558, 0.01723906, 0.92504, 0.9998292, 0.9996583, 0.02941151),
data.frame("CGGP1", 90, 2195, 0.03973626, -5.939358, 0.01805724, 0.94663, 0.9998205, 0.9996409, 0.0301442),
data.frame("CGGP1", 90, 2695, 0.03895927, -5.709819, 0.01752450, 0.89132, 0.9998274, 0.9996548, 0.02955428),
data.frame("CGGP", 90, 399, 0.03292523, -6.292347, 0.01476177, 0.93929, 0.9998769, 0.9997534, 0.02502954),
data.frame("CGGP", 90, 599, 0.02445636, -6.717207, 0.01163083, 0.95573, 0.999932, 0.999864, 0.01859961),
data.frame("CGGP", 90, 799, 0.02426409, -6.831546, 0.01119061, 0.95034, 0.9999334, 0.9998661, 0.01841507),
data.frame("CGGP", 90, 999, 0.02418435, -6.846158, 0.01062984, 0.95242, 0.9999338, 0.999867, 0.01836773),
data.frame("CGGP", 90, 1199, 0.02205223, -7.085375, 0.009788555, 0.95352, 0.9999454, 0.9998894, 0.01662363),
data.frame("CGGP", 90, 1699, 0.01527591, -7.696946, 0.007269341, 0.97937, 0.9999735, 0.9999469, 0.01145086),
data.frame("CGGP", 90, 2199, 0.01370636, -7.730417, 0.006900497, 0.9812, 0.9999787, 0.9999573, 0.01036107),
data.frame("CGGP", 90, 2699, 0.014988 , -7.711674, 0.006816212, 0.98524, 0.9999745, 0.9999489, 0.01092403),
data.frame("CGGP", 90, 3199, 0.01357989, -8.056265, 0.005935432, 0.98193, 0.9999791, 0.9999581, 0.009891518),
data.frame("CGGP", 90, 4199, 0.01130804, -8.147965, 0.005525203, 0.97767, 0.9999855, 0.9999709, 0.008477671),
data.frame("CGGP", 90, 5199, 0.009739982,-8.454266, 0.004678459, 0.98001, 0.9999892, 0.9999784, 0.007307837),
data.frame("CGGP", 90, 6195, 0.007799561, -8.601915, 0.004069435, 0.99209, 0.9999931, 0.9999862, 0.005720459),
data.frame("CGGP", 90, 7187, 0.007602274, -8.832714, 0.003677643, 0.9891, 0.9999934, 0.9999869, 0.005552559),
data.frame("CGGP", 90, 8179, 0.006772034, -9.045547, 0.003310222, 0.98978, 0.9999948, 0.9999896, 0.004955794),
data.frame("CGGP", 90, 10163,0.005923366, -9.315209, 0.002928836, 0.98941, 0.999996, 0.999992, 0.004356485),
data.frame("CGGP", 90, 12159,0.004279414, -9.339384, 0.002790037, 0.99488, 0.9999979, 0.9999958, 0.003197408),
data.frame("CGGP", 90, 16159,0.003443654, -9.731616, 0.002267292, 0.99656, 0.9999987, 0.9999973, 0.002545515),
data.frame("CGGP", 90, 18159,0.003443654, -9.731616, 0.002267292, 0.99656, 0.9999987, 0.9999973, 0.002545515),
data.frame("CGGP", 90, 20155,0.003984703, -9.738379, 0.002263422, 0.99591, 0.9999982, 0.9999964, 0.002905657),
data.frame("CGGP", 90, 22155,0.004459119, -9.739725, 0.002274312, 0.99623, 0.9999977, 0.9999955, 0.00319555),
data.frame("CGGP", 90, 24155,0.004439284, -9.703938, 0.002317367, 0.99618, 0.9999978, 0.9999955, 0.003193208),
data.frame("CGGP", 90, 26155,0.004961369, -9.659113, 0.002383479, 0.99554, 0.9999972, 0.9999944, 0.00355455),
data.frame("CGGPIgnore", 90, 26155,0.004959593, -9.656147, 0.002386668, 0.99554, 0.9999972, 0.9999944, 0.003553383),
data.frame("CGGPMat32", 90, 26155, 0.004838821, -8.49521 , 0.003830926, 0.99942, 0.9999973, 0.9999947, 0.003493466),
data.frame("CGGP18ktheta", 90, 26155, 0.004768937, -7.572385, 0.005782846, 0.99923, 0.9999974, 0.9999948, 0.003410945),
data.frame("CGGP18kthetanosupp", 90, 26155, 0.00473138, -7.165235, 0.007330487, 0.99932, 0.9999975, 0.9999949, 0.003383729),
# data.frame("CGGP", 90, 2199, ),
# data.frame("CGGP", 90, , ),
# data.frame("CGGP", 90, , ),
data.frame("CGGP3", 90, 399, 0.03086891, -6.507128, 0.01359139, 0.9464, 0.9998919, 0.9997833, 0.023462),
data.frame("CGGP3", 90, 599, 0.02465403, -6.811887, 0.01158544, 0.94892, 0.999931, 0.9998618, 0.01870552),
data.frame("CGGP3", 90, 799, 0.02501423, -6.701828, 0.01198723, 0.9419, 0.9999293, 0.9998577, 0.01896552),
data.frame("CGGP3", 90, 999, 0.02635975, -6.529451, 0.01207771, 0.90545, 0.9999212, 0.999842 , 0.02004152),
data.frame("CGGP3", 90, 1199, 0.02474022, -6.754248, 0.01105155, 0.90884, 0.9999306, 0.9998608, 0.01878582),
data.frame("CGGP3", 90, 1699, 0.01494981, -7.536539, 0.007537477, 0.97898, 0.9999747, 0.9999492, 0.01117594),
data.frame("CGGP3", 90, 2199, 0.01467389, -7.641863, 0.00727886, 0.97162, 0.9999755, 0.999951 , 0.01102318),
data.frame("CGGP3", 90, 2699, 0.01491939, -7.712265, 0.006820505, 0.98566, 0.9999747, 0.9999494, 0.01086603),
data.frame("CGGP3", 90, 3199, 0.01306396, -8.091254, 0.005814433, 0.98243, 0.9999806, 0.9999612, 0.009469989),
data.frame("CGGP3", 90, 4199, 0.01082473, -8.154173, 0.005511459, 0.98191, 0.9999867, 0.9999734, 0.008157038),
data.frame("CGGP3", 90, 5199, 0.00822481, -8.572066, 0.004325123, 0.98704, 0.9999923, 0.9999846, 0.006089218),
data.frame("CGGP3", 90, 6195, 0.008069555, -8.623623, 0.004043901, 0.99013, 0.9999926, 0.9999852, 0.005929349),
data.frame("CGGP3", 90, 7187, 0.007608375, -8.91085 , 0.003599635, 0.98793, 0.9999934, 0.9999868, 0.005580803),
data.frame("CGGP3", 90, 8179, 0.00736447, -9.108104, 0.003316512, 0.98517, 0.9999938, 0.9999877, 0.005399971),
data.frame("CGGP3", 90, 10163, 0.006123244, -9.453641, 0.002815331, 0.98794, 0.9999958, 0.9999915, 0.004529484),
data.frame("CGGP3", 90, 12163, 0.004061737, -9.652619, 0.002411071, 0.99395, 0.9999981, 0.9999962, 0.003012229),
data.frame("CGGP3", 90, 16163, 0.004127308, -9.924129, 0.002172679, 0.99046, 0.9999981, 0.9999961, 0.003106536),
data.frame("CGGP3", 90, 18163, 0.003674585, -10.05135, 0.002041532, 0.99177, 0.9999985, 0.9999969, 0.002787224),
data.frame("CGGP3", 90, 20159, 0.003491477, -10.04612, 0.001995089, 0.99532, 0.9999986, 0.9999972, 0.002585274),
data.frame("CGGP3", 90, 30155, 0.004722018, -10.10524, 0.00193077, 0.99323, 0.9999975, 0.9999949, 0.003389804),
data.frame("CGGP3PE",90,20159, 0.003089233, -8.211388, 0.004281498, 1, 0.9999989, 0.9999978, 0.002299494),
data.frame("CGGP3PE",90,30155, 0.003431799, -8.39904, 0.003919711, 0.99961, 0.9999987, 0.9999973, 0.002488821),
data.frame("CGGP3C" ,90,20159, 0.003050064, -8.897864, 0.003109479, 0.99982, 0.9999989, 0.9999979, 0.002269927),
data.frame("CGGP3C" ,90,30155, 0.003797589, -8.963827, 0.003034772, 0.99897, 0.9999984, 0.9999967, 0.002741234),
# data.frame("CGGP3", 90, , ),
# mlegp
data.frame("mlegp", 0, 50, 0.2121086, -2.136166, 0.1122491, 0.9941, 0.9949418, 0.9897676, 0.1614444),
data.frame("mlegp", 0, 75, 0.1266577, -2.708455, 0.0773198, 0.99936, 0.9982067, 0.9963514, 0.09525388),
data.frame("mlegp", 0,100, 0.1005995, -2.896368, 0.06750803, 0.9998, 0.9988543, 0.9976983, 0.07583394),
data.frame("mlegp", 0,150, 0.08580729,-4.263628, 0.04071472, 0.83153, 0.9991757, 0.9983254, 0.06421618),
data.frame("mlegp", 0,200, 0.06383321, -4.68457, 0.02916151, 0.79007, 0.9995385, 0.9990733, 0.04804422),
data.frame("mlegp", 0,250, 0.05941459,-5.003207, 0.02610797, 0.80917, 0.9996011, 0.9991971, 0.04498041)
)
allstats <- lapply(allstats, function(x){colnames(x) <- c("Package", 'Nsup',"Ngrid","RMSE","score","CRPscore","coverage","corr","R2","RMSEnorm");x})
allstats <- do.call(rbind, allstats)
allstats$Ntotal <- allstats$Nsup + allstats$Ngrid
allstats$Package <- as.character(allstats$Package)
library(ggplot2)
# ggplot(data=allstats, mapping=aes(Ntotal, RMSE, color=Package, shape=as.factor(Nsup))) + geom_point(size=3) + scale_y_log10()
ggplot(data=allstats, mapping=aes(Ntotal, RMSE, color=Package, shape=as.factor(Nsup))) + geom_point(size=3) + scale_x_log10() + scale_y_log10()
ggplot(data=allstats, mapping=aes(Ntotal, RMSE, color=Nsup)) + geom_point(size=3) + facet_grid(. ~ Package) + scale_x_log10() + scale_y_log10()
ggplot(data=allstats, mapping=aes(Ntotal, score, color=(Package), shape=as.factor(Nsup))) + geom_point(size=3) + scale_x_log10()
ggplot(data=allstats, mapping=aes(Ntotal, CRPscore, color=(Package), shape=as.factor(Nsup))) + geom_point(size=3) + scale_x_log10() + scale_y_log10()
# ggplot(data=allstats %>% filter(score<1e5), mapping=aes(Ntotal, score, color=(Package), shape=as.factor(Nsup))) + geom_point(size=3) + scale_x_log10()
tdf <- rbind(cbind(Ngrid=1699, stats.rt.sggp.1699),
cbind(Ngrid=2195, stats.rt.sggp.2195),
cbind(Ngrid=2695, stats.rt.sggp.2695))
ggplot(data=tdf, mapping=aes(Ngrid, RMSE)) + geom_point()
# Plot for thesis
tdf2 <- allstats %>% filter(Package %in% c("CGGP3", "mlegp"))
tdf2$Package[tdf2$Package == "CGGP3"] <- "CGGP"
ggrmse <- ggplot(data=tdf2, mapping=aes(Ntotal, RMSE, color=Package, shape=Package)) + geom_point(size=4) + scale_x_log10() + scale_y_log10() + xlab("Number of points evaluated"); ggrmse
maybe_save("redTimeRMSE", ggrmse, height=6, width=7)
# Plots for dissertation for specific redTime object
r3.sggp.20159 <- readRDS("./scratch/redTime/redTimeData/out_Big3_SGGP-20159.rds")
# blocks plot
r3.sggp.20159 %>% CGGPplotblocks() + coord_fixed()
maybe_save("redTime_plotblocks", r3.sggp.20159 %>% CGGPplotblocks() + coord_fixed(), height=6, width=6)
# Correlation plot
r3.sggp.20159 %>% CGGPplotcorr()
maybe_save("redTime_corr", r3.sggp.20159 %>% CGGPplotcorr(), height=8, width=6.8)
# Projection plot
r3.sggp.20159 %>% CGGPplotprojection(outdims = c(20,50,80))
r3.sggp.20159 %>% CGGPplotprojection(outdims = c(50))
r3.sggp.20159 %>% CGGPplotprojection(outdims = c(50), facet="wrap")
maybe_save("redTimeProjection", r3.sggp.20159 %>% CGGPplotprojection(outdims = c(50), facet="wrap"), height=6, width=6)
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.