Thesis: Exploratory Data Analaysi
Gerhard Viljoen
11/25/2018
require(reticulate)Python code for creating a JSON file from Dictionary Objects
"""
This script extracts Python Dictionaries with Tracklet information
"""
import os
from ast import literal_eval
import json
os.chdir("/Users/gerhard/MSc-thesis/data")
dat_files = os.listdir("/Users/gerhard/MSc-thesis/data")
dat_files.pop(16)
for i in range(0,len(dat_files)):
print(dat_files[i])
d = open(dat_files[i])
d = d.read()
d = literal_eval(d)
jayson = json.dumps(d,indent=4,sort_keys=True)
name1="dat"
name2=".json"
name=name1+str(i)+name2
outfile = open(name,"w")
outfile.write(jayson)Plotting energy deposit in a single detector pad
As a heatmap
setwd("~/MSc-thesis/testdict")
require(jsonlite)
test.dict <- fromJSON("test.json")
an.image <- test.dict$`0`$layer0
image(an.image)
As a filled contour map
filled.contour(an.image,color.palette = heat.colors)
filled.contour(x=1:11,y=1:24,z=unlist(an.image),color.palette = heat.colors,levels=0:30)
Plotting a detector pad’s energy deposit, next to the transpose of the same matrix
par(mfrow=c(2,2))
L=2
for(i in 1:L){
a <- test.dict[[i]]$layer0
if(is.null(a)==F & is.null(ncol(a))==F & is.null(nrow(a))==F){
image(x=1:11,y=1:24,a)
a <- t(a)
image(x=1:24,y=1:11,a)
# filled.contour(a,color.palette = heat.colors,plot.title = title(paste0("Event: ",test.dict[[i]]$Event," TrackID: ",test.dict[[i]]$V0TrackID," PDG Code:",
# test.dict[[i]]$pdgCode)))
}
}
Plot 3D histograms
Unscaled 3D Surface Plot
require(plotly)
require(plot3D)
a <- test.dict[[1]]$layer0
plotly::plot_ly(z=~a) %>% add_surface()Normalized and standardized 3D Surface Plot
plotly::plot_ly(z=~scale(a)) %>% add_surface()Three examples of pion energy deposition 3D histograms
L <- length(test.dict)
count.pion <- 0
count.electron <- 0
for(i in 1:L){
a <- test.dict[[i]]$layer0
if(is.null(a)==F & is.null(ncol(a))==F & is.null(nrow(a))==F & sum(a)!=0 & count.pion < 3){
if(test.dict[[i]]$pdgCode==211 | test.dict[[i]]$pdgCode==-211){
count.pion <- count.pion+1
hist3D(z=a)
}
}
}
Three examples of electron energy deposition 3D Histograms
for(i in 1:L){
a <- test.dict[[i]]$layer0
if(is.null(a)==F & is.null(ncol(a))==F & is.null(nrow(a))==F & count.electron < 3){
if(test.dict[[i]]$pdgCode==11 | test.dict[[i]]$pdgCode==-11){
count.electron <- count.electron + 1
hist3D(z=a)
}
}
}
Read in 100 python dictionaries
(memory intensive so commented out here)
# rm(list=ls())
# setwd("~/MSc-thesis/data")
# require(jsonlite)
# require(readtext)
#
# #PDG codes
# pdg.elec <- c(11,-11)
# pdg.pion <- c(-211,211)
#
# files <- list.files(path="~/MSc-thesis/data", pattern="*json", full.names=T, recursive=FALSE)
#
# j <- fromJSON(files[1])
#
# for(i in 2:length(files)){
#
# f <- fromJSON(files[i])
# j <- c(j,f)
# }Split combined JSON out into 2 separate JSON objects for protons & electrons
(memory intensive so commented out here)
Electrons
# electrons <- numeric(264)
#
# L <- length(j)
#
# for(i in 1:L){
#
# a <- j[[i]]$layer0
#
#
# if(is.null(a)==F & is.null(ncol(a))==F & is.null(nrow(a))==F & sum(unlist(a)) != 0){
#
#
# if(j[[i]]$pdgCode==11 | j[[i]]$pdgCode==-11){
# a <- as.vector(j[[i]]$layer0)
# electrons <- rbind(electrons,a)
#
# }
#
# }
#
# }
#
# electrons <- as.data.frame(electrons)
# electrons <- electrons[-1,]
# write.csv(electrons,"electrons100.csv")Pions
# pions <- numeric(264)
#
# L <- length(j)
#
# for(i in 1:L){
#
# a <- j[[i]]$layer0
#
# if(is.null(a)==F & is.null(ncol(a))==F & is.null(nrow(a))==F & sum(unlist(a)) != 0){
#
#
# if(j[[i]]$pdgCode==211 | j[[i]]$pdgCode==-211){
# a <- as.vector(j[[i]]$layer0)
# pions <- rbind(pions,a)
#
# }
#
# }
#
# }
#
# pions <- as.data.frame(pions)
# pions <- pions[-1,]
#
# write.csv(pions,"pions100.csv")Clean up environment
rm(list=ls())
e <- read.csv("electrons100.csv")
p <- read.csv("pions100.csv")Some high-level stats
Distribution of energy deposition in a detector pad:
Pions are plotted in blue, electrons in red
e <- e[,-1]
mean.electron.e.dep.pad <- rowMeans(e)
p <- p[,-1]
mean.pion.e.dep.pad <- rowMeans(p)
plot(density(mean.pion.e.dep.pad),col="blue",main="Blue = Pions; Red = Electrons",sub= "Electron v Pion: Energy deposit per pad distribution",xlab = "Average energy deposition per PID: dashed lines")
lines(density(mean.electron.e.dep.pad),col="red")
abline(v=mean(mean.pion.e.dep.pad),lty="dashed",col="blue")
abline(v=mean(mean.electron.e.dep.pad),lty="dashed",col="red")
Averaged pad data
Electron
electron.pad <- numeric(264)
for(i in 1:nrow(e)){
this.pad <- unlist(e[i,])
electron.pad <- electron.pad + this.pad
}
electron.pad <- matrix(electron.pad,nrow=24,byrow=T)
filled.contour(electron.pad,color.palette = heat.colors)
Pion
pion.pad <- numeric(264)
for(i in 1:nrow(p)){
this.pad <- unlist(p[i,])
pion.pad <- pion.pad + this.pad
}
pion.pad <- matrix(pion.pad,nrow=24,byrow=T)
filled.contour(pion.pad,color.palette = heat.colors)
filled.contour(scale(electron.pad),color.palette = heat.colors)
filled.contour(scale(pion.pad),color.palette = heat.colors)
Plot 100 electron traces:
par(mfrow=c(2,5))
for(i in 1:100){
this.pad <- unlist(e[i,])
this.pad <- matrix(this.pad,nrow=24,byrow=T)
image(this.pad)
}
Plot 100 pion traces:
par(mfrow=c(2,5))
for(i in 1:100){
this.pad <- unlist(p[i,])
this.pad <- matrix(this.pad,nrow=24,byrow=T)
image(this.pad)
}
Try to figure out the path a particle takes
By fitting a linear regression line through a 3D scatterplot of pad data
require(plotly)
z <- unlist(e[1,])
y <- rep(1:24,11)
x <- rep(1:11,24)
dat <- cbind(x,y,z)
dat <- data.frame(dat)
names(dat) <- c("x","y","z")
linmod <- lm(z~x+y,dat)
l <- predict(linmod,dat)
l <- matrix(l,nrow=24,byrow = T)
plot_ly(dat,x=~x,y=~y,z=~z) %>% add_markers() %>%
add_surface(z=~l)require(plotly)
z <- unlist(e[1,])
y <- rep(1:24,11)
x <- rep(1:11,24)
dat <- cbind(x,y,z)
dat <- data.frame(dat)
names(dat) <- c("x","y","z")
linmod <- lm(z~x+y+x:y,dat)
l <- predict(linmod,dat)
l <- matrix(l,nrow=24,byrow = T)
plot_ly(dat,x=~x,y=~y,z=~z) %>% add_markers() %>%
add_surface(z=~l)