README

Combining package nanotime for operating with nanosecond time-resolution with package data.table leverages the conciseness, high performance, and memory efficiency of the latter to provide high-resolution, high-performance time series operations.

Our time-series representation is simply a data.table with a first column of type nanotime and a key on it. This means all the standard data.table functions can be used, and this package consolidates this functionality.

Specifically, dtts proposes alignment functions that are particularly versatile, and allow to work across time-zones.

Usage

Creating a data.table-based time-series with a nanotime index

Three operations are necessary to create a data.table-based time-series for use with the functions defined in this package: 1. Create the time index, i.e. a vector of nanotime 2. Create a data.table with the first column being the time index and specifying it as a key

For instance, this code creates a time-series of 10 rows spaced every hour with a data column V1 containing random data:

library(data.table)
library(nanotime)
t1 <- seq(as.nanotime(Sys.time()), by=as.nanoduration("01:00:00"), length.out=10)
dt1 <- data.table(index=t1, V1=runif(10), key="index")

dt1 <- data.table(index = seq(as.nanotime(Sys.time()), by=as.nanoduration("01:00:00"), length.out=10),
                  V1 = runif(10),
                  key = "index")

Alignment functions

Alignment is the process of matching the time of the observations of one time series to another. All alignment functions in this package work in a similar way. For each point in the vector y onto which x is aligned, a pair or arguments named start and end define an interval around this point. As an example let us take start equal to -1 hour and end equal to 0 hour. This means that a y of 2021-11-20 11:00:00 defines an interval from 2021-11-20 10:00:00 to 2021-11-20 11:00:00. The alignment process will then use that interval to pick points in order to compute one or more statistics on that interval for the corresponding point in y.

In addition to the arguments start and end, two other arguments, booleans named sopen and eopen, define if the start and end, respectively, of the interval are open or not.

Finally, note that when the interval is specified with a nanoperiod type, the argument tz is necessary in order to give meaning to the interval. With nanoperiod, alignments are time-zone aware and correct across daylight saving time.

This figure shows an alignment using a statistic (here simply counting the number of elements in the intervals):

align_idx

This function takes two vectors of type nanotime. It aligns the first one onto the second one and returns the indices of the first vector that align with the second vector. There is no choice of aggregation function here as this function works uniquely on nanotime vectors. The algorithm selects the point in x that falls in the interval that is closest to the point of alignment in y. The index of the point that falls in that interval is returned at the position of the vector y. If no point exists in that interval NaN is returned.

library(dtts)

t1 <- seq(as.nanotime("1970-01-01T00:00:00+00:00"), by=as.nanoduration("00:00:01"), length.out=100)
t2 <- seq(as.nanotime("1970-01-01T00:00:10+00:00"), by=as.nanoduration("00:00:10"), length.out=10)

align_idx(t1, t2, start=as.nanoduration("-00:00:10"))

align

This function takes a data.table and aligns it onto y, a vector of nanotime. Like align_idx, it uses the arguments start, end, sopen and eopen to define the intervals around the points in y.

Instead of the result being an index, it is a new data.table time-series with the first nanotime column being the vector y, and the rows of this time-series are taken from the data.table x. If no function is specified (i.e. func is NULL), the function returns the row of the point in x that is in the interval and that is closest to the point in y on which the alignment is made. If func is defined, it receives for each point in y all the rows in x that are in the defined interval. So func must be a statistic that returns one row, but it may return one or more columns. Common examples are means (e.g. using colMeans), counts, etc.

In the following example a time-series dt1 is created with a data column V1 which has the integer index as value and it is aligned onto a nanotime vector t2

library(dtts)

t1 <- seq(as.nanotime("1970-01-01T00:00:00+00:00"), by=as.nanoduration("00:00:01"), length.out=100)
dt1 <- data.table(index=t1, V1=0:99)
setkey(dt1, index)

t2 <- seq(as.nanotime("1970-01-01T00:00:10+00:00"), by=as.nanoduration("00:00:10"), length.out=10)

align(dt1, t2, start=as.nanoduration("-00:00:10"), func=colMeans)

grid_align

This function adds one more dimension to the function align. Instead of taking a vector y, it constructs a grid that has as interval the value supplied in the argument by. The interval is controllable (with arguments ival_start, ival_end, ival_sopen, ival_eopen) but it is likely that in most cases the default will be used which is the grid interval. As in the case of align, the caller can specify func. Finally, note that by can be either a nanoduration or a nanoperiod. In the latter case, as for the other functions, the argument tz must be supplied so that the nanoperiod interval can be anchored to a specific timezone.

The following example is the same as for the align function, but shows that the vector t2 does not need to be supplied as it is instead constructed by grid_align:

library(dtts)

t1 <- seq(as.nanotime("1970-01-01T00:00:00+00:00"), by=as.nanoduration("00:00:01"), length.out=100)
dt1 <- data.table(index=t1, V1=0:99)
setkey(dt1, index)

grid_align(dt1, as.nanoduration("00:00:10"), func=colMeans)

Frequency

Using grid_align and nrow it is possible to get the frequency of a time-series, i.e. to count the number of elements in each interval of a grid.

Taking the same example as above, we see that the result is the count of elements of dt1 that are in each interval:

library(dtts)

t1 <- seq(as.nanotime("1970-01-01T00:00:00+00:00"), by=as.nanoduration("00:00:01"), length.out=100)
dt1 <- data.table(index=t1, V1=0:99)
setkey(dt1, index)

grid_align(dt1, as.nanoduration("00:00:10"), func=nrow)

ops

ops performs arithmetic operations between two time-series and has the following signature, where x and y are time-series and op is a string denoting an arithmetic operator.

ops(x, y, op_string)

Each entry in the left time-series operand defines an interval from the previous entry, and the value associated with this interval will be applied to all the observations in the right time-series operand that fall in the interval. Note that the interval is closed at the beginning and open and the end. The available values for op are “*“,”/“,”+“,”-“.

This function is particulary useful to apply a multiplier or to add a constant that changes over time; one example would be the adjustment of stock prices for splits.

one_second_duration  <- as.nanoduration("00:00:01")
t1 <- nanotime(1:2 * one_second_duration * 3)
t2 <- nanotime(1:4 * one_second_duration)
dt1 <- data.table(index=t1, data1 = 1:length(t1))
setkey(dt1, index)
dt2 <- data.table(index=t2, data1 = 1:length(t2))
setkey(dt2, index)
ops(dt1, dt2, "+")

Time-series subsetting

one_second <- 1e9
index <- seq(nanotime("2022-12-12 12:12:10+00:00"), length.out=10, by=one_second)
dts <- data.table(index=index, data=1:length(index), key="index")
ival <- as.nanoival(c("-2022-12-12 12:12:10+00:00 -> 2022-12-12 12:12:14+00:00-"),
                     ("+2022-12-12 12:12:18+00:00 -> 2022-12-12 12:12:20+00:00+"))
dts[index %in% ival]

Status

dtts currently proposes only a set of alignment functions, but it is likely that other time-series functions will be impletemented so that nanotime-based time-series have reasonably complete time-series functionality.

See the issue tickets for an up to date list of potentially desirable, possibly planned, or at least discussed items.

Installation

install.packages("dtts")

remotes::install_github("eddelbuettel/dtts")

dtts: Time-series functionality based on `nanotime` and `data.table`.

Motivation

Usage

Creating a `data.table`-based time-series with a `nanotime` index

Alignment functions

`align_idx`

`align`

`grid_align`

Frequency

ops

Time-series subsetting

Status

Installation

Author

License

dtts: Time-series functionality based on nanotime and data.table.

Motivation

Usage

Creating a data.table-based time-series with a nanotime index

Alignment functions

align_idx

align

grid_align

Frequency

ops

Time-series subsetting

Status

Installation

Author

License

dtts: Time-series functionality based on `nanotime` and `data.table`.

Creating a `data.table`-based time-series with a `nanotime` index

`align_idx`

`align`

`grid_align`