Synchronise and plot two timeseries data sets in R - r

Question

I have two data sets from an experiment on a person during different ambient temperatures. P1 represents a patient's physiological response data and P1IAQ represent the environmental monitoring data during the experiment. P1 recorded data 32 times per second but P1IAQ recorded data every 10 seconds.
head(P1IAQ)
Time RH Temp CO2
1 12:04:07 44.2 19.89 664
2 12:04:17 44.2 19.89 664
3 12:04:27 44.2 19.89 665
4 12:04:37 44.2 19.89 665
5 12:04:47 44.2 19.89 666
6 12:04:57 44.2 19.89 668
head(P1)
Time SkinTemp HeartRate RespirationRate
1 00:00:00 27.781 70 10
2 00:00:00 27.780 70 10
3 00:00:00 27.779 70 10
4 00:00:00 27.779 70 10
5 00:00:00 27.778 70 10
6 00:00:00 27.777 70 10
The problem I have is that the time stamp on P1 is wrong. How can I plot them together on the same graph to see if the SkinTemp has a time-lag after the environmental temp is decreased?
EDIT: dput for P1IAQ
I've added the first twenty values for the environmental data. I think the best thing would be to subtract 12:04:07 from all values to make the starting time 00:00:00. I've tried looking at lubridate.
library(lubridate)
P1IAQ$Time<-hms(P1IAQ$Time)
This datetime post looks interesting, but it's for plotting dates rather than actually altering them.
Plotting data against time in R
dput(P1IAQ)
structure(list(Time = structure(1:19, .Label = c("12:04:07",
"12:04:17", "12:04:27", "12:04:37", "12:04:47", "12:04:57", "12:05:07",
"12:05:17", "12:05:27", "12:05:37", "12:05:47", "12:05:57", "12:06:07",
"12:06:17", "12:06:27", "12:06:37", "12:06:47", "12:06:57", "12:07:07"
), class = "factor"), RH = c(44.2, 44.2, 44.2, 44.2, 44.2, 44.2,
44.2, 44.2, 44.1, 44.1, 44.2, 44.2, 44.2, 44.3, 44.2, 44.2, 44.2,
44.3, 44.3), Temp = c(19.89, 19.89, 19.89, 19.89, 19.89, 19.89,
19.89, 19.89, 19.89, 19.89, 19.94, 19.89, 19.94, 19.94, 19.94,
19.94, 19.94, 19.94, 19.94), CO2 = c(664L, 664L, 665L, 665L,
666L, 668L, 668L, 669L, 667L, 670L, 670L, 672L, 675L, 677L, 682L,
684L, 685L, 686L, 687L)), .Names = c("Time", "RH", "Temp", "CO2"
), class = "data.frame", row.names = c(NA, -19L))
EDIT: I've synchronised the times using lubridate:
P1IAQ$Time<-period_to_seconds(hms(as.character(P1IAQ$Time))-hms("12:04:07"))
P1$Time<-period_to_seconds(hms(as.character(P1$Time)))
But now plotting them together is tricky. I've tried ggplot2 but I can't get two vertical axes. Any thoughts
ggplot() +
geom_line(data = P1IAQ, aes(x = Time, y = Temp, color = "red")) +
geom_line(data = P1, aes(x = Time, y = Temp, color = "blue")) +
xlab('Time (s)') +
ylab('Temperature ºC')

Answer 1

If you only need the times in each data frame to be on a common scale, you can convert both of them to numeric seconds elapsed since the start of the experiment and not worry about date or time classes. Then you can join the two data frames based on the common time scale.
I used your P1IAQ data sample and created fake P1 data to go with it. Time in my P1 is probably not in the same format as your actual data. If you post a sample of your P1, I can adjust the example below to fit your actual data.
library(dplyr)
library(reshape2)
library(hms)
library(zoo)
library(ggplot2)
theme_set(theme_light())
# Fake P1 data frame
set.seed(10)
n=32*60*3 + 1
P1 = data.frame(Time=as.POSIXct(seq(0,180,length.out=n), origin=as.Date("2016-05-01"), tz="GMT"),
SkinTemp = round(cumsum(rnorm(n, 0, 0.01)) + 27.78, 2),
RespirationRate=round(rnorm(n, 10, 0.5)))
Convert P1$Time and P1IAQ$Time to numeric values equal to the number of seconds elapsed since the start of the experiment. (Note that P1IAQ$Time in the data you posted is a factor, so I converted to character before further processing.):
P1$nTime = as.numeric(as.hms(P1$Time))
P1IAQ$nTime = as.numeric(as.hms(as.character(P1IAQ$Time)))
P1IAQ$nTime = P1IAQ$nTime - min(P1IAQ$nTime)
Join P1 and P1IAQ by nTime:
P1j = full_join(P1, P1IAQ, by="nTime", suffix=c("_P1","_P1IAQ")) %>%
# Make sure joined data frame is sorted by nTime
arrange(nTime) %>%
# Fill missing values with Last One Carried Forward
mutate_at(vars(Time_P1IAQ, RH, Temp, CO2), na.locf)
Plot after converting data from wide to long format:
ggplot(P1j %>% select(Time_P1IAQ, nTime, Skin=SkinTemp, Ambient=Temp) %>%
# Convert from wide to long format for plotting
melt(id.var=c("Time_P1IAQ", "nTime")),
aes(nTime, value, group=Time_P1IAQ)) +
geom_line() +
facet_grid(variable ~ ., scales="free_y") +
scale_y_continuous(expand=c(0.5,0)) +
labs(x="Elapsed Time (sec)", y=expression(Temperature~"("*degree*C*")"))
Another option is to plot temperature changes relative to the start of the experiment. That way, you can have both lines on the same panel without having to deal with them being in different locations:
ggplot(P1j %>% select(Time_P1IAQ, nTime, Skin=SkinTemp, Ambient=Temp) %>%
# Convert from wide to long format for plotting
melt(id.var=c("Time_P1IAQ", "nTime")) %>%
# Convert temperatures to difference from starting values
group_by(variable) %>%
mutate(value = value - value[nTime==min(nTime)]),
aes(nTime, value, colour=variable)) +
geom_line() +
labs(x="Elapsed Time (sec)", y=expression(Temperature~Change~"("*degree*C*")"),
colour="")

Answer 2

I don't have your data but i will prepare something similar... In this case the length of P1 is diferent with P1IAQ:
library(ggplot2)
#I create a sample of your data
P1<-data.frame(1:10,51:60)
P1IAQ<-data.frame(1:8,1:8)
colnames(P1)<-c("Time","Temp")
colnames(P1IAQ)<-c("Time","Temp")
# I cathegory your data for plot
df = data.frame(Time=c(P1$Time,P1IAQ$Time), values=c(P1$Temp,P1IAQ$Temp),type=c(rep("P1",length(P1$Time)),rep("P1IAQ",length(P1IAQ$Time))))
ggplot(data=df, aes(x=Time, y=values, color=type)) +
geom_line() +
facet_grid(type ~ ., scales="free") +
xlab('Time (s)') +
ylab('Temperature ºC')