Time period buckets - r

I've got logs of events that contain:
start time, end time, category id and count. They cover several months.
I'd like to aggregate them over time to be able to trace histograms over a given day, week, month.
So I assume the best way to do this is to bin the periods in buckets. I think 5 minutes would be good.
e.g. If an event starts at 1.01pm and ends at 1.07pm, I'd like to obtain 2 records for it as it covers 2 periods of 5 minutes (0-5 and 5-10) and replicate the rest of the original data for these new records (category and count)
if my input logs (x) are as such:
start / end / catid / count
2012-11-17 15:05:02.0, 2012-11-17 15:12:52.0, 1, 2
2012-11-17 15:07:13.0, 2012-11-17 15:17:47.0, 2, 10
2012-11-17 15:11:00.0, 2012-11-17 15:12:33.0, 3, 5
2012-11-17 15:12:01.0, 2012-11-17 15:20:00.0, 4, 1
I'm trying to get the output bucketed in 5 minutes (b) this way:
start / catid / count
2012-11-17 15:05:00.0 1, 2
2012-11-17 15:10:00.0 1, 2
2012-11-17 15:05:00.0 2, 10
2012-11-17 15:10:00.0 2, 10
2012-11-17 15:15:00.0 2, 10
2012-11-17 15:10:00.0 3, 5
2012-11-17 15:10:00.0 4, 1
2012-11-17 15:15:00.0 4, 1
Then I can easily aggregate the new data frame (b) over category ids for the period I want (hour, day, week, month)
I'm starting with R and I found a lot explanations on how to bucket a time value but not a period of time.
I've had a look at zoo and xts but I couldn't quite find what to do.
Hopefully that makes sense to some of you.
Edit:
I've slightly modified Ram's suggestion to get the correct calculation of blocks using the rounded endtime rather than the original end time. (Thanks Ram!)
mnslot=15 # size of the buckets/slot in minutes
#Round down the minutes of starttime to a mutliple of mnslot
st.str <- strptime(st, "%Y-%m-%d %H:%M:%S")
min_st <- as.numeric(format(st.str, "%M"))
roundedmins <- floor(min_st/mnslot) * mnslot
st.base <- strptime(st, "%Y-%m-%d %H")
rounded_start <- st.base + (roundedmins * 60)
#Round down the minutes of the endtime to a multiple of mnslot.
en.str <- strptime(en, "%Y-%m-%d %H:%M:%S")
min_en <- as.numeric(format(en.str, "%M"))
roundedmins <- floor(min_en/mnslot) * mnslot
en.base <- strptime(en, "%Y-%m-%d %H")
rounded_end<- en.base + (roundedmins * 60)
# calculate the number of blocks based on the rounded minutes of start and end
numblocks<- as.numeric(floor((rounded_end-rounded_start)/mnslot/60)+1)
# differenced of POSIXct values is in minutes
# but difference of POSIXlt seems to be in seconds , so have to divide by 60 as well
#Create REPLICATED Rows, depending on the size of the interval
replicated_cat = NULL
replicated_count = NULL
replicated_start = NULL
for (n in 1:length(numblocks)){
for (newrow in 1:numblocks[n]){
replicated_start = c(replicated_start, df$rounded_start[n]+(newrow-1)*300 )
replicated_cat = c(replicated_cat, df$catid[n])
replicated_count = c(replicated_count, df$count[n])
}
}
#Change to readable format
POSIXT <- unix2POSIXct(replicated_start)
newdf <- data.frame(POSIXT, replicated_cat, replicated_count)
names(newdf) <- c("start", "CatId", "Count")
newdf
This produces the required output. it is a bit slow though:p

Here's a fully working version. It involves step-by-step data manipulation for what you are after.
#storing the original data as a csv
df <- read.csv("tsdata.csv")
st<-as.POSIXlt(df$start)
en<-as.POSIXlt(df$end)
#a utility function to convert formats
unix2POSIXct <- function (time) structure(time, class = c("POSIXt", "POSIXct") )
#For each row, determine how many replications are needed
numdups <- as.numeric(floor((en-st)/5)+1)
st.str <- strptime(st, "%Y-%m-%d %H:%M:%S")
min_st <- as.numeric(format(st.str, "%M"))
#Round down the minutes of start to 5 minute starts. 0,5,10 etc...
roundedmins <- floor(min_st/5) * 5
st.base <- strptime(st, "%Y-%m-%d %H")
df$rounded_start <- st.base + (roundedmins * 60)
#Create REPLICATED Rows, depending on the size of the interval
replicated_cat = NULL
replicated_count = NULL
replicated_start = NULL
for (n in 1:length(numdups)){
for (newrow in 1:numdups[n]){
replicated_start = c(replicated_start, df$rounded_start[n]+(newrow-1)*300 )
replicated_cat = c(replicated_cat, df$catid[n])
replicated_count = c(replicated_count, df$count[n])
}
}
#Change to readable format
POSIXT <- unix2POSIXct(replicated_start)
newdf <- data.frame(POSIXT, replicated_cat, replicated_count)
names(newdf) <- c("start", "CatId", "Count")
newdf
Which produces:
start CatId Count
1 2012-11-17 15:05:00 1 2
2 2012-11-17 15:10:00 1 2
3 2012-11-17 15:05:00 2 10
4 2012-11-17 15:10:00 2 10
5 2012-11-17 15:15:00 2 10
6 2012-11-17 15:10:00 3 5
7 2012-11-17 15:10:00 4 1
8 2012-11-17 15:15:00 4 1

That's not an easy one ... I am also missing the structure of the whole problem so I hope it is ok if I limit myself to outlining the basic approach, if things are unclear you can come back to me.
First (if I were you) I would install the 'lubridate' package, which makes playing around with dates/times a lot easier.
Then maybe try something like this:
z <- strptime("17/11/12 15:05:00.0", "%d/%m/%y %H:%M:%OS")
This will define your starting point in time, if that is supposed to be defined by the first logs(x) time then there is the minute command available e.g.
z <- strptime("17/11/12 15:05:02.0", "%d/%m/%y %H:%M:%OS")
minute(z)<-5;second(z)<-0.0 #I guess, you get the concept
Then produce a sequence of 5 minute intervals
z5s<-z+minutes(seq(0,100,5))
This will produce a sequence of 20, 5 minute time intervals, here again I do not know how flexible the whole thing is supposed to be.
Finally you could then play around with for instance modulo operations
z2<-z+minutes(2)
z2 should be the end time, I just added 2 minutes "manually" here to illustrate the concept
(as.integer(z2-z))%%5 > 5
FALSE
or if you want to see how many 5 minute spans are covered only do (as.integer(z2-z))%%5
or whatever other functions you prefer to match/distribute the log times across the z5s POSIXlt intervals.
Hope this helps a bit i.e. gives you some direction.

Related

Create 10,000 date data.frames with fake years based on 365 days window

Here my time period range:
start_day = as.Date('1974-01-01', format = '%Y-%m-%d')
end_day = as.Date('2014-12-21', format = '%Y-%m-%d')
df = as.data.frame(seq(from = start_day, to = end_day, by = 'day'))
colnames(df) = 'date'
I need to created 10,000 data.frames with different fake years of 365days each one. This means that each of the 10,000 data.frames needs to have different start and end of year.
In total df has got 14,965 days which, divided by 365 days = 41 years. In other words, df needs to be grouped 10,000 times differently by 41 years (of 365 days each one).
The start of each year has to be random, so it can be 1974-10-03, 1974-08-30, 1976-01-03, etc... and the remaining dates at the end df need to be recycled with the starting one.
The grouped fake years need to appear in a 3rd col of the data.frames.
I would put all the data.frames into a list but I don't know how to create the function which generates 10,000 different year's start dates and subsequently group each data.frame with a 365 days window 41 times.
Can anyone help me?
#gringer gave a good answer but it solved only 90% of the problem:
dates.df <- data.frame(replicate(10000, seq(sample(df$date, 1),
length.out=365, by="day"),
simplify=FALSE))
colnames(dates.df) <- 1:10000
What I need is 10,000 columns with 14,965 rows made by dates taken from df which need to be eventually recycled when reaching the end of df.
I tried to change length.out = 14965 but R does not recycle the dates.
Another option could be to change length.out = 1 and eventually add the remaining df rows for each column by maintaining the same order:
dates.df <- data.frame(replicate(10000, seq(sample(df$date, 1),
length.out=1, by="day"),
simplify=FALSE))
colnames(dates.df) <- 1:10000
How can I add the remaining df rows to each col?
The seq method also works if the to argument is unspecified, so it can be used to generate a specific number of days starting at a particular date:
> seq(from=df$date[20], length.out=10, by="day")
[1] "1974-01-20" "1974-01-21" "1974-01-22" "1974-01-23" "1974-01-24"
[6] "1974-01-25" "1974-01-26" "1974-01-27" "1974-01-28" "1974-01-29"
When used in combination with replicate and sample, I think this will give what you want in a list:
> replicate(2,seq(sample(df$date, 1), length.out=10, by="day"), simplify=FALSE)
[[1]]
[1] "1985-07-24" "1985-07-25" "1985-07-26" "1985-07-27" "1985-07-28"
[6] "1985-07-29" "1985-07-30" "1985-07-31" "1985-08-01" "1985-08-02"
[[2]]
[1] "2012-10-13" "2012-10-14" "2012-10-15" "2012-10-16" "2012-10-17"
[6] "2012-10-18" "2012-10-19" "2012-10-20" "2012-10-21" "2012-10-22"
Without the simplify=FALSE argument, it produces an array of integers (i.e. R's internal representation of dates), which is a bit trickier to convert back to dates. A slightly more convoluted way to do this is and produce Date output is to use data.frame on the unsimplified replicate result. Here's an example that will produce a 10,000-column data frame with 365 dates in each column (takes about 5s to generate on my computer):
dates.df <- data.frame(replicate(10000, seq(sample(df$date, 1),
length.out=365, by="day"),
simplify=FALSE));
colnames(dates.df) <- 1:10000;
> dates.df[1:5,1:5];
1 2 3 4 5
1 1988-09-06 1996-05-30 1987-07-09 1974-01-15 1992-03-07
2 1988-09-07 1996-05-31 1987-07-10 1974-01-16 1992-03-08
3 1988-09-08 1996-06-01 1987-07-11 1974-01-17 1992-03-09
4 1988-09-09 1996-06-02 1987-07-12 1974-01-18 1992-03-10
5 1988-09-10 1996-06-03 1987-07-13 1974-01-19 1992-03-11
To get the date wraparound working, a slight modification can be made to the original data frame, pasting a copy of itself on the end:
df <- as.data.frame(c(seq(from = start_day, to = end_day, by = 'day'),
seq(from = start_day, to = end_day, by = 'day')));
colnames(df) <- "date";
This is easier to code for downstream; the alternative being a double seq for each result column with additional calculations for the start/end and if statements to deal with boundary cases.
Now instead of doing date arithmetic, the result columns subset from the original data frame (where the arithmetic is already done). Starting with one date in the first half of the frame and choosing the next 14965 values. I'm using nrow(df)/2 instead for a more generic code:
dates.df <-
as.data.frame(lapply(sample.int(nrow(df)/2, 10000),
function(startPos){
df$date[startPos:(startPos+nrow(df)/2-1)];
}));
colnames(dates.df) <- 1:10000;
>dates.df[c(1:5,(nrow(dates.df)-5):nrow(dates.df)),1:5];
1 2 3 4 5
1 1988-10-21 1999-10-18 2009-04-06 2009-01-08 1988-12-28
2 1988-10-22 1999-10-19 2009-04-07 2009-01-09 1988-12-29
3 1988-10-23 1999-10-20 2009-04-08 2009-01-10 1988-12-30
4 1988-10-24 1999-10-21 2009-04-09 2009-01-11 1988-12-31
5 1988-10-25 1999-10-22 2009-04-10 2009-01-12 1989-01-01
14960 1988-10-15 1999-10-12 2009-03-31 2009-01-02 1988-12-22
14961 1988-10-16 1999-10-13 2009-04-01 2009-01-03 1988-12-23
14962 1988-10-17 1999-10-14 2009-04-02 2009-01-04 1988-12-24
14963 1988-10-18 1999-10-15 2009-04-03 2009-01-05 1988-12-25
14964 1988-10-19 1999-10-16 2009-04-04 2009-01-06 1988-12-26
14965 1988-10-20 1999-10-17 2009-04-05 2009-01-07 1988-12-27
This takes a bit less time now, presumably because the date values have been pre-caclulated.
Try this one, using subsetting instead:
start_day = as.Date('1974-01-01', format = '%Y-%m-%d')
end_day = as.Date('2014-12-21', format = '%Y-%m-%d')
date_vec <- seq.Date(from=start_day, to=end_day, by="day")
Now, I create a vector long enough so that I can use easy subsetting later on:
date_vec2 <- rep(date_vec,2)
Now, create the random start dates for 100 instances (replace this with 10000 for your application):
random_starts <- sample(1:14965, 100)
Now, create a list of dates by simply subsetting date_vec2 with your desired length:
dates <- lapply(random_starts, function(x) date_vec2[x:(x+14964)])
date_df <- data.frame(dates)
names(date_df) <- 1:100
date_df[1:5,1:5]
1 2 3 4 5
1 1997-05-05 2011-12-10 1978-11-11 1980-09-16 1989-07-24
2 1997-05-06 2011-12-11 1978-11-12 1980-09-17 1989-07-25
3 1997-05-07 2011-12-12 1978-11-13 1980-09-18 1989-07-26
4 1997-05-08 2011-12-13 1978-11-14 1980-09-19 1989-07-27
5 1997-05-09 2011-12-14 1978-11-15 1980-09-20 1989-07-28

Count rows based on multiple consecutive time flows

I have a large file of time-series data, which looks as follows. The dataset covers years, in increments of 15 minutes. A small subset looks like:
uniqueid time
a 2014-04-30 23:30:00
a 2014-04-30 23:45:00
a 2014-05-01 00:00:00
a 2014-05-01 00:15:00
a 2014-05-12 13:45:00
a 2014-05-12 14:00:00
b 2014-05-12 13:45:00
b 2014-05-12 14:00:00
b 2014-05-12 14:30:00
To reproduce above:
time<-c("2014-04-30 23:30:00","2014-04-30 23:45:00","2014-05-01 00:00:00","2014-05-01 00:15:00",
"2014-05-12 13:45:00","2014-05-12 14:00:00","2014-05-12 13:45:00","2014-05-12 14:00:00",
"2014-05-12 14:30:00")
uniqueid<-c("a","a","a","a","a","a","b","b","b")
mydf<-data.frame(uniqueid,time)
My goal is to count the number of rows per unique id, per consecutive timeflow. A consecutive timespan is when a unique id is stamped for each 15 minutes in a row (such as id A, which is stamped from 30.04.14 23.30 hrs until 01.05.14 00.15 hrs - hence 4 rows), yet when this flow of 15-minute iterations is disrupted (after 01.05.14 00:15, it is not stamped at 01.05.14 00:30 hence it is disrupted), it should count the next timestamp as start of a new consecutive timeflow and again calculate the number of rows until this flow is disrupted again. Time is POSIX.
As you can see in above example; a consecutive timeflow may cover different days, different months, or different years. I have many unique ids (and as said, a very large file), so I'm looking for a way that my computer can handle (loops probably wouldn't work).
I am looking for output something like:
uniqueid flow number_rows
a 1 4
a 2 2
b 3 2
b 4 1
I have looked into some time packages (such as lubridate), but given my limited R knowledge, I don't even know where to begin.
I hope all is clear - if not, I'd be happy to try to clarify it further. Thank you very much in advance!
Another way to do this with data.table also using a time difference would be to make use of the data.table internal values for group number and number of rows in each group:
library(data.table)
res<-setDT(mydf)[, list(number_rows=.N,flow=.GRP),
by=.(uniqueid,cumsum(as.numeric(difftime(time,shift(time,1L,type="lag",fill=0))) - 15))][,cumsum:=NULL]
print(res)
uniqueid number_rows flow
1: a 4 1
2: a 2 2
3: b 2 3
4: b 1 4
Also since the sample data you posted didn't align with the subset you posted, I have included my data below:
Data
time<-as.POSIXct(c("2014-04-30 23:30:00","2014-04-30 23:45:00","2014-05-01 00:00:00","2014-05-01 00:15:00",
"2014-05-12 13:45:00","2014-05-12 14:00:00","2014-05-12 13:45:00","2014-05-12 14:00:00",
"2014-05-12 14:30:00"))
uniqueid<-c("a","a","a","a","a","a","b","b","b")
mydf<-data.frame(uniqueid,time)
You can groupby the uniqueid and the cumulative sum of the difference of time between rows which is not equal to 15 min and that gives the flow id and then a count of rows should give you what you need:
A justification of the logic is whenever the time difference is not equal to 15 within each uniqueid, a new flow process should be generated so we label it as TRUE and combine that with the cumsum, it becomes a new flow id with the following consecutive rows:
library(dplyr)
mydf$time <- as.POSIXct(mydf$time, "%Y-%m-%d %H:%M:%S")
# convert the time column to POSIXct class so that we can apply the diff function correctly
mydf %>% group_by(uniqueid, flow = 1 + cumsum(c(F, diff(time) != 15))) %>%
summarize(num_rows = n())
# Source: local data frame [4 x 3]
# Groups: uniqueid [?]
#
# uniqueid flow num_rows
# <fctr> <dbl> <int>
# 1 a 1 4
# 2 a 2 2
# 3 b 3 2
# 4 b 4 1
Base R is pretty fast. Using crude benchmarking, I found it finished in half the time of DT, and I got tired of waiting for dplyr.
# estimated size of data, years x days x hours x 15mins x uniqueids
5*365*24*4*1000 # = approx 180M
# make data with posixct and characters of 180M rows, mydf is approx 2.5GB in memory
time<-rep(as.POSIXct(c("2014-04-30 23:30:00","2014-04-30 23:45:00","2014-05-01 00:00:00","2014-05-01 00:15:00",
"2014-05-12 13:45:00","2014-05-12 14:00:00","2014-05-12 13:45:00","2014-05-12 14:00:00",
"2014-05-12 14:30:00")),times = 20000000)
uniqueid<-rep(as.character(c("a","a","a","a","a","a","b","b","b")),times = 20000000)
mydf<-data.frame(uniqueid,time = time)
rm(time,uniqueid);gc()
Base R:
# assumes that uniqueid's are in groups and in order, and there won't be a followed by b that have the 15 minute "flow"
starttime <- Sys.time()
# find failed flows
mydf$diff <- c(0,diff(mydf$time))
mydf$flowstop <- mydf$diff != 15
# give each flow an id
mydf$flowid <- cumsum(mydf$flowstop)
# clean up vars
mydf$time <- mydf$diff <- mydf$flowstop <- NULL
# find flow length
mydfrle <- rle(mydf$flowid)
# get uniqueid/flowid pairs (unique() is too slow)
mydf <- mydf[!duplicated(mydf$flowid), ]
# append rle and remove separate var
mydf$number_rows <- mydfrle$lengths
rm(mydfrle)
print(Sys.time()-starttime)
# Time difference of 30.39437 secs
data.table:
library(data.table)
starttime <- Sys.time()
res<-setDT(mydf)[, list(number_rows=.N,flow=.GRP),
by=.(uniqueid,cumsum(as.numeric(difftime(time,shift(time,1L,type="lag",fill=0))) - 15))][,cumsum:=NULL]
print(Sys.time()-starttime)
# Time difference of 57.08156 secs
dplyr:
library(dplyr)
# convert the time column to POSIXct class so that we can apply the diff function correctly
starttime <- Sys.time()
mydf %>% group_by(uniqueid, flow = 1 + cumsum(c(F, diff(time) != 15))) %>%
summarize(num_rows = n())
print(Sys.time()-starttime)
# too long, did not finish after a few minutes
I think the assumption of uniqueid's and times being in order is huge, and the other solutions might be able to take advantage of that better. order() is easy enough to do.
I'm not sure about the impact of memory, or of the impact of different data sets that aren't so simple. It should be easy enough to break it into chunks and process if memory is an issue. It takes more code in Base R for sure.
Having both ordered "id" and "time" columns, we could build a single group to operate on by creating a logical vector of indices wherever either "id" changes or "time" is > 15 minutes.
With:
id = as.character(mydf$uniqueid)
tm = mydf$time
find where "id":
id_gr = c(TRUE, id[-1] != id[-length(id)])
and "time":
tm_gr = c(0, difftime(tm[-1], tm[-length(tm)], unit = "mins")) > 15
change and combine them in:
gr = id_gr | tm_gr
which shows wherever either "id" changed or "time" > 15.
And to get the result:
tab = tabulate(cumsum(gr)) ## basically, the only operation per group -- 'n by group'
data.frame(id = id[gr], flow = seq_along(tab), n = tab)
# id flow n
#1 a 1 4
#2 a 2 2
#3 b 3 2
#4 b 4 1
On a larger scale:
set.seed(1821); nid = 1e4
dat = replicate(nid, as.POSIXct("2016-07-07 12:00:00 EEST") +
cumsum(sample(c(1, 5, 10, 15, 20, 30, 45, 60, 90, 120, 150, 200, 250, 300), sample(5e2:1e3, 1), TRUE)*60),
simplify = FALSE)
names(dat) = make.unique(rep_len(letters, nid))
dat = data.frame(id = rep(names(dat), lengths(dat)), time = do.call(c, dat))
system.time({
id = as.character(dat$id); tm = dat$time
id_gr = c(TRUE, id[-1] != id[-length(id)])
tm_gr = c(0, difftime(tm[-1], tm[-length(tm)], unit = "mins")) > 15
gr = id_gr | tm_gr
tab = tabulate(cumsum(gr))
ans1 = data.frame(id = id[gr], flow = seq_along(tab), n = tab)
})
# user system elapsed
# 1.44 0.19 1.66
For comparison, included MikeyMike's answer:
library(data.table)
dat2 = copy(dat)
system.time({
ans2 = setDT(dat2)[, list(flow = .GRP, n = .N),
by = .(id, cumsum(as.numeric(difftime(time,
shift(time, 1L, type = "lag", fill = 0),
unit = "mins")) > 15))][, cumsum := NULL]
})
# user system elapsed
# 3.95 0.22 4.26
identical(as.data.table(ans1), ans2)
#[1] TRUE

Data frame of departure and return dates, how do I get a list of all dates away?

I'm stuck on a problem calculating travel dates. I have a data frame of departure dates and return dates.
Departure Return
1 7/6/13 8/3/13
2 7/6/13 8/3/13
3 6/28/13 8/7/13
I want to create and pass a function that will take these dates and form a list of all the days away. I can do this individually by turning each column into dates.
## Turn the departure and return dates into a readable format
Dept <- as.Date(travelDates$Dept, format = "%m/%d/%y")
Retn <- as.Date(travelDates$Retn, format = "%m/%d/%y")
travel_dates <- na.omit(data.frame(dept_dates,retn_dates))
seq(from = travel_dates[1,1], to = travel_dates[1,2], by = 1)
This gives me [1] "2013-07-06" "2013-07-07"... and so on. I want to scale to cover the whole data frame, but my attempts have failed.
Here's one that I thought might work.
days_abroad <- data.frame()
get_days <- function(x,y){
all_days <- seq(from = x, to = y, by =1)
c(days_abroad, all_days)
return(days_abroad)
}
get_days(travel_dates$dept_dates, travel_dates$retn_dates)
I get this error:
Error in seq.Date(from = x, to = y, by = 1) : 'from' must be of length 1
There's probably a lot wrong with this, but what I would really like help on is how to run multiple dates through seq().
Sorry, if this is simple (I'm still learning to think in r) and sorry too for any breaches in etiquette. Thank you.
EDIT: updated as per OP comment.
How about this:
travel_dates[] <- lapply(travel_dates, as.Date, format="%m/%d/%y")
dts <- with(travel_dates, mapply(seq, Departure, Return, by="1 day"))
This produces a list with as many items as you had rows in your initial table. You can then summarize (this will be data.frame with the number of times a date showed up):
data.frame(count=sort(table(Reduce(append, dts)), decreasing=T))
# count
# 2013-07-06 3
# 2013-07-07 3
# 2013-07-08 3
# 2013-07-09 3
# ...
OLD CODE:
The following gets the #days of each trip, rather than a list with the dates.
transform(travel_dates, days_away=Return - Departure + 1)
Which produces:
# Departure Return days_away
# 1 2013-07-06 2013-08-03 29 days
# 2 2013-07-06 2013-08-03 29 days
# 3 2013-06-28 2013-08-07 41 days
If you want to put days_away in a separate list, that is trivial, though it seems more useful to have it as an additional column to your data frame.

R: Efficiently subsetting dataframe based on time of day

I have a large (150,000x7) dataframe that I intend to use for back-testing and real-time analysis of a financial market. The data represents the condition of an investment vehicle at 5 minute intervals (although holes do exist). It looks like this (but much longer):
pTime Time Price M1 M2 M3 M4
1 1212108300 20:45:00 1.5518 12.21849 -0.37125 4.50549 -31.00559
2 1212108900 20:55:00 1.5516 11.75350 -0.81792 -1.53846 -32.12291
3 1212109200 21:00:00 1.5512 10.75070 -1.47438 -8.24176 -34.35754
4 1212109500 21:05:00 1.5514 10.23529 -1.06044 -8.46154 -33.24022
5 1212109800 21:10:00 1.5514 9.74790 -1.02759 -10.21978 -33.24022
6 1212110100 21:15:00 1.5513 9.31092 -1.17076 -11.97802 -33.79888
7 1212110400 21:20:00 1.5512 8.84034 -1.28428 -13.62637 -34.35754
8 1212110700 21:25:00 1.5509 8.07843 -1.63715 -18.24176 -36.03352
9 1212111000 21:30:00 1.5509 7.39496 -1.49198 -20.65934 -36.03352
10 1212111300 21:35:00 1.5512 7.65266 -1.03717 -18.57143 -34.35754
The data is pre-loaded into R, but during my back-test I need to subset it by two criteria:
The first criteria is a sliding window to avoid peeking into the future. The window must be such that, each new 5 minute interval on the back-test shifts the whole window into the future by 5 minutes. This part I can do like this:
require(zoo)
zooser <- zoo(x=tser$Close, order.by=as.POSIXct(tser$pTime, origin="1970-01-01"))
window(zooser, start=A, end=B)
The second criteria is another sliding window, but one that slides through time of day and contains only those entries that are within N minutes of the input time on any given day.
Example: If the window's size is 2 hours, and the input time is 12:00PM then the window must contain all rows with Time between 10:00AM and 2:00PM
This is the part that I am having trouble figuring out.
Edit: My data has holes in it, two consecutive rows could be MORE than 5 minutes apart. The data looks like this (very zoomed in)
As the window moves through these gaps the number of points inside the windows should vary.
The following is my MySQL code that does what I want to do in R (same table structure):
SET #qTime = Time(FROM_UNIXTIME(SAMP_endTime));
SET #inc = -1;
INSERT INTO MetIndListBuys (pTime,ArrayPos,M1,M2,M3,M4)
SELECT pTime,#inc:=#inc+1,M1,M2,M3,M4
FROM mergebuys USE INDEX (`y`) WHERE pTime BETWEEN SAMP_startTime AND SAMP_endTime
AND TIME_TO_SEC(TIMEDIFF(Time,#qTime))/3600 BETWEEN 0-HourSpan AND HourSpan
;
Say that you have your target time t0 on the same scale as pTime: seconds since epoch. Then t0 - pTime = (difference in the number of days since epoch between the two) + (difference in remaining seconds). Taking t0 - pTime %% (num. seconds per day) will leave us with the difference in seconds in clock arithmetic (wrapped around if the difference is negative). This suggests the following function:
SecondsPerDay <- 24 * 60 * 60
within <- function(d, t0Sec, wMin) {
diff <- (d$pTime - t0Sec) %% SecondsPerDay
wSec <- 60 * wMin
return(d[diff < wSec | diff > (SecondsPerDay - wSec), ])
}
1) If DF is the data frame shown in the question then create a zoo object from it as you have done and split it into days giving zs. Then lapply your function f to each successive set of w points in each component (i.e. in each day). For example, if you want to apply your function to 2 hours of data at a time and your data is regularly spaced 5 minute data then w = 24 (since there are 24 five minute periods in two hours). In such a case f would be passed 24 rows of data as a matrix each time its called. Also align has been set to "right" below but it can alternately be set to align="center" and the condition giving ix can be changed to double sided, etc. For more on rollapply see: ?rollapply
library(zoo)
z <- zoo(DF[-2], as.POSIXct(DF[,1], origin = "1970-01-01"))
w <- 3 # replace this with 24 to handle two hours at a time with five min data
f <- function(x) {
tt <- x[, 1]
ix <- tt[w] - tt <= w * 5 * 60 # RHS converts w to seconds
x <- x[ix, -1]
sum(x) # replace sum with your function
}
out <- rollapply(z, w, f, by.column = FALSE, align = "right")
Using the data frame in the question we get this:
> out
$`2008-05-30`
2008-05-30 02:00:00 2008-05-30 02:05:00 2008-05-30 02:10:00 2008-05-30 02:15:00
-66.04703 -83.92148 -95.93558 -100.24924
2008-05-30 02:20:00 2008-05-30 02:25:00 2008-05-30 02:30:00 2008-05-30 02:35:00
-108.15038 -121.24519 -134.39873 -140.28436
By the way, be sure to read this post .
2) This could alternately be done as the following where w and f are as above:
n <- nrow(DF)
m <- as.matrix(DF[-2])
sapply(w:n, function(i) { m <- m[seq(length = w, to = i), ]; f(m) })
Replace the sapply with lapply if needed. Also this may seem shorter than the first solution but its not much different once you add the code to define f and w (which appear in the first but not the second).
If there are no holes during the day and only holes between days then these solutions could be simplified.

How to select and plot hourly averages from data frame?

I have a CSV file that looks like this, where "time" is a UNIX timestamp:
time,count
1300162432,5
1299849832,0
1300006132,1
1300245532,4
1299932932,1
1300089232,1
1299776632,9
1299703432,14
... and so on
I am reading it into R and converting the time column into POSIXct like so:
data <- read.csv(file="data.csv",head=TRUE,sep=",")
data[,1] <- as.POSIXct(data[,1], origin="1970-01-01")
Great so far, but now I would like to build a histogram with each bin corresponding to the average hourly count. I'm stuck on selecting by hour and then counting. I've looked through ?POSIXt and ?cut.POSIXt, but if the answer is in there, I am not seeing it.
Any help would be appreciated.
Here is one way:
R> lines <- "time,count
1300162432,5
1299849832,0
1300006132,1
1300245532,4
1299932932,1
1300089232,1
1299776632,9
1299703432,14"
R> con <- textConnection(lines); df <- read.csv(con); close(con)
R> df$time <- as.POSIXct(df$time, origin="1970-01-01")
R> df$hour <- as.POSIXlt(df$time)$hour
R> df
time count hour
1 2011-03-15 05:13:52 5 5
2 2011-03-11 13:23:52 0 13
3 2011-03-13 09:48:52 1 9
4 2011-03-16 04:18:52 4 4
5 2011-03-12 12:28:52 1 12
6 2011-03-14 08:53:52 1 8
7 2011-03-10 17:03:52 9 17
8 2011-03-09 20:43:52 14 20
R> tapply(df$count, df$hour, FUN=mean)
4 5 8 9 12 13 17 20
4 5 1 1 1 0 9 14
R>
Your data doesn't actually yet have multiple entries per hour-of-the-day but this would average over the hours, properly parsed from the POSIX time stamps. You can adjust with TZ info as needed.
You can calculate the hour "bin" for each time by converting to a POSIXlt and subtracting away the minute and seconds components. Then you can add a new column to your data frame that would contain the hour bin marker, like so:
date.to.hour <- function (vec)
{
as.POSIXct(
sapply(
vec,
function (x)
{
lt = as.POSIXlt(x)
x - 60*lt$min - lt$sec
}),
tz="GMT",
origin="1970-01-01")
}
data$hour <- date.to.hour(as.POSIXct(data[,1], origin="1970-01-01"))
There's a good post on this topic on Mages' blog. To get the bucketed data:
aggregate(. ~ cut(time, 'hours'), data, mean)
If you just want a quick graph, ggplot2 is your friend:
qplot(cut(time, "hours"), count, data=data, stat='summary', fun.y='mean')
Unfortunately, because cut returns a factor, the x axis won't work properly. You may want to write your own, less awkward bucketing function for time, e.g.
timebucket = function(x, bucketsize = 1,
units = c("secs", "mins", "hours", "days", "weeks")) {
secs = as.numeric(as.difftime(bucketsize, units=units[1]), units="secs")
structure(floor(as.numeric(x) / secs) * secs, class=c('POSIXt','POSIXct'))
}
qplot(timebucket(time, units="hours"), ...)

Resources