Time Series and Signal Aggregation

Description

Aggregation and coursening of time series and signals. This is the method for the aggregate function for objects of class timeSeries and signalSeries.

Usage

aggregateSeries(x, pos, FUN, moving=FALSE, together=FALSE, drop.empty=TRUE,
                include.ends=FALSE, adj, offset, colnames, by, k.by=1,
                week.align=NULL, holidays=timeDate(), align.by=TRUE,
                incr=1, ...)

Arguments

Value

returns a new time series whose positions are the adjusted passed-in positions or positions generated from by, k.by, and so on, (or possibly a subset if drop.empty is TRUE) and whose rows are aggregated from x as specified in the arguments. Aggregation takes place by separating x into blocks separated by the positions (or overlapping blocks with a fixed number of samples if moving is supplied), and then applying FUN to each column (or all columns together if together is TRUE) and forming a new time series with the positions and aggregated data.

See Also

timeSeries, signalSeries, align, aggregate

Examples

x <- timeSeries(data.frame(1:20,rnorm(20)), timeCalendar(d=1:20))
aggregate(x, FUN=mean, by="weeks")

Time Series and Signal Interpolation and Alignment

Description

Aligns or interpolates a time series or signal to new positions.

Usage

align(x, pos, how="NA", error.how="NA", localzone=FALSE, matchtol=0, by, 
      k.by=1, week.align=NULL, holidays=timeDate())

Arguments

Details

If either x or pos (or the generated sequence) has zero length, a zero-length series is returned.

Value

returns a new time series or a signal whose positions are the passed-in positions or positions generated from by, k.by, and so on, and whose rows are derived from x as specified in the arguments. (Can be a subset if how or error.how is "drop".)

See Also

timeSeries, signalSeries, positions, seriesMerge.

Examples

a <- signalSeries(pos=1:10, data=data.frame(a = 11:20, b = 5 * (1:10))) 
align(a, c(.2, 3, 7.8, 12), how = "interp", error.how = "nearest") 
a <- timeSeries(pos=as(1:10, "timeDate"), 
		data=data.frame(a = 11:20, b = 5 * (1:10))) 
alpos <- as(c(.2, 3, 7.8, 12), "timeDate") 
alpos@time.zone <- "JST" 
positions(a)@time.zone <- "PST" 
align(a, alpos, matchtol = 1, localzone = TRUE) 
align(a, matchtol=1, localzone=TRUE, by="days", k.by=2) 

Uniform Rectangular Data Functions

Description

Functions that allow you to access all rectangular data objects in the same way. Rectangular data objects include matrices, data frames and vectors.

Usage

as.rectangular(x) 
as.char.rect(x) 
is.rectangular(x) 
subscript2d(x,i,j) 
subscript2d(x,i,j) <- value
numRows(x) 
numRows(x) <- value
numCols(x) 
numCols(x) <- value
rowIds(x) 
rowIds(x) <- value
colIds(x) 
colIds(x) <- value

Arguments

Details

subscript2d, numRows, numCols, rowIds, colIds can also be used on the left side of assignments. The value can be a character vector, or anything that can be coerced to a character vector.

• subscript2d is for subscripting. When subscript2d is used in an assignment, it does not allow subscript replacement outside the bounds of x. Instead, set numRows or numCols first.

• When numRows or numCols is used in an assignment, the row and column IDs are maintained to have the correct length. Usually, this is done by setting numRows on the ID vector, but for some objects (for example, data frames) this might not be appropriate, and they have their own methods.

• Functions colnames<- and rownames<- simply call colIds<- and rowIds<-, respectively.

• as.rectangular converts any object to a rectangular data object (usually a data frame), if possible.

• is.rectangular tests whether an object is rectangular.

• numRows and numCols count the number of rows and columns.

• rowIds and colIds (and rownames and colnames) return the row and column names or other identifiers attached to rows and columns.

• colnames and rownames return the same values as colIds and rowIds, respectively, if do.NULL=T.

• Instead of using names to replace row names from a matrix, use rowIds or dimnames.

• The functions colnames, rownames, colnames<-, rownames<- emulate R functions of the same names.

Value

See Also

as.data.frame, matrix, Subscript, nrow, dimnames.

Examples

x <- 1:10 
y <- list(a=1:10, b=11:20) 
is.rectangular(x) 
y <- as.rectangular(y) 
subscript2d(x,3,1) 
subscript2d(y,4,1) <- 55 
numRows(x) 
numCols(y) <- 3 
rowIds(x) <- letters[1:10] 
colIds(y) 
z <- cbind(1,1:4)
colnames(z)
colnames(z) <- colnames(z)
rownames(z) <- rownames(z)

Dow Jones Industrial Average

Description

High, low, opening, and closing prices and trading volume for the Dow Jones Industrial Average.

The data set has:

• The closing price only from 1915 through September 1928.

• The high, low, and closing prices from October 1928 through March 9, 1984.

• The high, low, opening, and closing prices from March 12, 1984 through December 1986.

• The high, low, opening, and closing prices and the trading volume from January 1987 through February 1990.

Format

An object of class timeSeries with the high, low, open and close prices stored as a data.frame.

Source

Downloaded from the College of Business, Ohio State University web site in early 1990.

Foreign Exchange Rates

Description

Exchange rates between the US Dollar and the

• British Pound (GBP)

• Canadian Dollar (CAD)

• German Mark (DEM)

• Japanese Yen (JPY)

• Swiss Franc (CHF)

in a multi-variate time series.

Data from Andreas S. Weigend, Bernardo A. Huberman, and David E. Rumelhart, "Predicting Sunspots and Exchange Rates with Connectionist Networks", pp. 395-432 in M. Casdagli and S. Eubank, eds, Nonlinear Modeling and Forecasting, Addison-Wesley, 1992.

Federal Reserve Interest Rates

Description

Interest rate data from the web site of the Federal Reserve Bank, https://www.federalreserve.gov/releases/h15/data.htm, running from 1972 to 1997.

You can find more information on that web site.

Documentation below is derived from the data files.

fed.rate is a multi-variate time series, with the following columns:

Value

High, Low, Open, and Close Calculation

Description

Calculates the high, low, first, and last elements of a vector. Especially useful for financial trading data in conjunction with the aggregateSeries function.

Usage

hloc(x)

Arguments

Value

returns a vector with four elements:

x can be an array, but dimensions are ignored.

See Also

aggregateSeries.

Examples

x <- c(5, 2, 3, 6, 3, 2, 1, 7, 1) 
hloc(x) 

Network Packet Traffic

Description

Time, type, and size for 10,000 network packets on a local intranet, just before 5PM on March 27, 1998, as reported by the Unix snoop command. Size is available only for packets whose type is "TCP" .

Format

An object of class timeSeries with the high, low, open and close prices stored as a data.frame.

Positions of series Objects

Description

Accesses the positions of series objects.

Usage

positions(object)
positions(object) <- value

Arguments

Details

This function can also be used on the left side of an assignment to set the positions of a series object.

Value

returns the positions slot of object.

See Also

seriesData, timeSeries, signalSeries.

Examples

x <- signalSeries(pos=1:10, data=11:20) 
positions(x) 
positions(x) <- 11:20 

Speech Signal

Description

A signal of a voice saying the word "wavelet" , sampled at 11025 Hz. There are 8192 samples, ranging in time from 0 to approximately 0.7429 seconds.

Format

The signal is of class signalSeries .

Base Class for Time Series and Signals

Description

A base class representing ordered data objects, such as time series and signals, that have positions (x values, times), and for each position a set of variables (stored in any rectangular data object).

Details

The series class holds x positions and variable data. It is valid only when the lengths of the positions and data match, and when the data slot is a rectangular object.

seriesVirtual is a virtual class corresponding to series. All of the methods for series objects are defined on the corresponding virtual seriesVirtual class so they can be inherited easily by extending classes.

series has two built-in extending classes: timeSeries and signalSeries. series is not meant to be used directly. Instead, most users should use the signalSeries and timeSeries classes. Extending classes should include both series and seriesVirtual in their representations.

Slots

data

(ANY) the variable data, which can be any data object for which is.rectangular is TRUE, such as a data.frame, matrix, or atomic vector.

positions

(positions) the x values for the variables.

start.position

(positions) the starting x value.

end.position

(positions) the ending x value.

future.positions

(positions) future x values used for predictions.

units

(character) units for the data.

title

(character) title of the data set.

documentation

(character) user-supplied documentation.

attributes

(ANY) attributes slot for arbitrary use.

Series functions

• The series class has a validity function, seriesValid.

• The access functions positions and seriesData can access the positions and data in the object, and they can be used on the left side of assignments.

• There are also methods defined for series objects for the following functions:

  • nrow

  • ncol

  • start

  • end

  • subscripting

  • the standard rectangular data functions (see is.rectangular)

  • basic arithmetic.

See Also

timeSeries class, signalSeries class, is.rectangular.

SeriesData of series Objects

Description

Accesses the seriesData of series objects.

Usage

seriesData(object)
asSeriesData(object)
seriesData(object) <- value
seriesDataNew()
seriesDataValid(object)

Arguments

Details

This function can also be used on the left side of an assignment to set the seriesData of a series object.

Value

returns the seriesData slot of object.

See Also

positions, timeSeries, signalSeries.

Examples

x <- signalSeries(pos=1:10, data=11:20) 
seriesData(x) 
seriesData(x) <- 1:10

Time Series Lag/Lead Function

Description

Returns a lagged/leading timeSeries or signalSeries object.

Usage

seriesLag(X, k = 1, trim = FALSE, pad = NA)

Arguments

Details

The difference between shift and seriesLag is that the returned series of shift is shifted in time (position) while the returned series of seriesLag shifts the entire data slot but keeps the same time (position) intact. They all work for both timeSeries and signalSeries objects.

Value

returns a lagged or leading time (signal) series of the original data.

See Also

lag, shift .

Examples

x <- timeSeries(data=data.frame(x=1:10, y=11:20),
    from="7/4/2000", by="bizdays")
seriesLag(x, 1)
seriesLag(x, -1)

Length of a timeSeries

Description

Returns the length of a timeSeries; that is, it returns the number of positions in the timeSeries.

Usage

seriesLength(x)

Arguments

Value

returns the length of the timeSeries.

Note

This function is distinguished from the length function, which returns the number of series in the timeSeries object.

See Also

timeSeries class.

Examples

x <- timeSeries(data=data.frame(x=1:10, y=11:20), from="7/4/2000", by="bizdays")
seriesLength(x)
length(x)

Merging for Time Series and Signals

Description

Merges time series or signal objects, making a new object with all the columns of the input objects, and some or all of the rows, depending on how their positions match.

Usage

seriesMerge(x1, x2, ..., pos=positions(x1), how,
        error.how, localzone=FALSE, matchtol=0,  
        suffixes)

Arguments

Value

returns a new series object containing all the columns of all the inputs, and all the rows of all the inputs, according to the alignment methods described above.

See Also

timeSeries, signalSeries, positions, align, merge.

Examples

a <- signalSeries(pos=1:10, data=data.frame(a = 11:20, b = 5 * (1:10))) 
b <- signalSeries(pos=5:14, data=data.frame(a = 11:20, b = 5 * (1:10))) 
seriesMerge(a, b) 
a <- timeSeries(pos=as(1:10, "timeDate"), 
		data=data.frame(a = 11:20, b = 5 * (1:10))) 
b <- timeSeries(pos=as(5:14, "timeDate"), 
	        data=data.frame(a = 11:20, b = 5 * (1:10))) 
seriesMerge(a, b, pos="union") 

Create a Shifted Time Series

Description

Returns a time series like the input but shifted in time.

Usage

shift(x, k=1)

Arguments

Details

shift is a generic function. Its default method calls lag(x,-k).

shift also has a method for series objects, which works for both timeSeries and signalSeries objects.

• To align the times of several new-style time series, use seriesMerge.

• To align the times of several old-style time series, use ts.intersect or ts.union.

• To compute a lagged/leading series with same time position but shifted data slot, use seriesLag. (seriesLag also works for both timeSeries and signalSeries objects.)

Value

returns a time series with the same data as x, but with positions lagged by k steps.

Note

The shift function replaces the lag function, which illogically had the opposite sign of shifting. (The lag function has been retained only because it is used in other functions.)

See Also

seriesMerge, lag, lag.plot, ts.intersect, ts.union.

Examples

x <- signalSeries(data=data.frame(a=1:10, b=letters[1:10]), positions=1:10)
x5 <- shift(x,5)
seriesMerge(x, x5, pos="union")

Create a signalSeries object

Description

Creates an object of class signalSeries

Usage

signalSeries(data, positions., units, units.position, from = 1, by = 1)

Arguments

Value

an object of class "signalSeries". See the signalSeries class help file for the names and structure of the slots in the object.

See Also

signalSeries class.

Examples

signalSeries(pos=1:10 , data=1:10)
signalSeries(data=data.frame(x=1:10, y=11:20), from=2, by=2)

signalSeries Class

Description

Represents non-calendar time series and signal objects.

Details

The signalSeries class inherits from the series and seriesVirtualclasses. It has slots that hold x positions and variable data inherited from the series class.

A signalSeries object is valid only when the lengths of the positions and data match, the data is a rectangular object, and the positions slot holds a positionsNumeric object.

Slots

All slots except the last, units.position, come from the series object.

data

(ANY) the variable data, which can be any data object for which is.rectangular is TRUE, such as a data.frame, matrix, or atomic vector.

positions

(positions) the x values for the variables, which must be of type positionsNumeric.

start.position

(positions) the starting x value.

end.position

(positions) the ending x value.

future.positions

(positions) future x values used for predictions.

units

(character) the units for the data.

title

(character) the title of the data set.

documentation

(character) user-supplied documentation.

attributes

(ANY) the attributes slot for arbitrary use.

units.position

(character) the units for the positions slot.

signalSeries functions

You can create objects of class signalSeries using the new function, in which case they are set up to be empty. Alternatively, you can create objects of class signalSeries using the signalSeries function.

These objects can be subscripted and used in mathematical operations much like data frames or matrices.

See Also

series class, signalSeries, is.rectangular.

Treasury Bill Auction Rates

Description

Treasury Bill auction rate data running from 1980 to 1997 for 3 month, 6 month, and 1 year durations.

Format

Three separate timeSeries objects with Treasury Bill auction rates:

tbauc.3m

Average of interest rate bids accepted in regular treasury auctions of 13-week bills (also known as 3-month bills). Currently, the auctions are held each Monday for bills to be issued the ensuing Thursday, in the absence of holidays.

tbauc.6m

Average of interest rate bids accepted in regular treasury auctions of 26-week bills (also known as 6-month bills). Currently, the auctions are held each Monday for bills to be issued the ensuing Thursday, in the absence of holidays.

tbauc.1y

Average of interest rate bids accepted in regular treasury auctions of 52-week bills (also known as 1-year bills). Currently, the auctions are held at roughly monthly intervals.

Source

From the web site of the Federal Reserve Bank, https://www.federalreserve.gov/releases/h15/data.htm.

Treasury Bond Futures Trading Data

Description

Treasury Bond futures trading data: high and low prices over 20-minute intervals from January 7, 1994, to February 3, 1995.

It is used to illustrate a drop in bond prices that occurred in 1994.

Format

An object of class timeSeries with the high and low prices stored as a data.frame.

Source

Downloaded from the Fisher College of Business, Ohio State University web site in early 1996.

Treasury Constant Maturity Curve

Description

Treasury Constant Maturity Curve data running from 1982 to 1997. The Constant Maturity Curve data come from yield curves constructed by the U.S. Treasury Department from the yields of actively traded issues adjusted to constant maturities.

Format

tcm.curve is a multivariate timeSeries object with the following columns:

three.month

Three-month rate.

six.month

Six-month rate.

one.year

One-year rate.

two.year

Two-year rate.

three.year

Three-year rate.

five.year

Five-year rate.

seven.year

Seven-year rate.

ten.year

Ten-year rate.

twenty.year

Constructed from the 20-year Treasury department numbers, based on the 20-year bond through December 1986 (at which time the 20-year bond was discontinued), and from the new computation starting in October of 1993 based on outstanding bonds with approximately 20 years remaining to maturity. There is no data between 1987 and September 1992.

thirty.year

Thirty-year rate.

long.term

An unweighted average of rates on all outstanding bonds neither due nor callable in less than 10 years, also calculated by the Treasury Department.

Source

From the web site of the Federal Reserve Bank, https://www.federalreserve.gov/releases/h15/data.htm.

Create a timeSeries Object

Description

Creates an object of class timeSeries.

Usage

timeSeries(data, positions., units., from = timeCalendar(d = 1,
    m = 1, y = 1960), by = "days", k.by = 1, align.by = FALSE,
    week.align = NULL)

Arguments

Value

an object of class "timeSeries". See the timeSeries class help file for the names and structure of the slots in the object.

See Also

timeSeries class.

Examples

timeSeries(pos=timeCalendar(d=1:10), data=1:10)
timeSeries(data=data.frame(x=1:10, y=11:20), from="7/4/2000", by="bizdays")

Calendar Time Series Class

Description

Represents calendar time series objects.

Details

The timeSeries class inherits series and seriesVirtual. From series, it inherits slots that hold x positions and variable data. A timeSeries object is valid only when the lengths of the positions and data match, the data slot is rectangular, and the positions slot holds a positionsCalendar object.

Slots

All slots except the last two, fiscal.year.start and type, are inherited from the base series class.

data

(ANY) the variable data, which can be any data object for which is.rectangular is TRUE, such as a data.frame, matrix, or atomic vector.

positions

(positions) the x values for the variables, which must be of type positionsCalendar.

start.position

(positions) the starting x value.

end.position

(positions) the ending x value.

future.positions

(positions) future x values used for predictions.

units

(character) the units for the data.

title

(character) the title of the data set.

documentation

(character) user-supplied documentation.

attributes

(ANY) the attributes slot for arbitrary use.

fiscal.year.start

(numeric) the month number for fiscal year start.

type

(character) the type of time series.

Series functions

You can create objects of class timeSeries using the new function, in which case they are set up to be empty and have their fiscal year starting in January. Alternatively, you can create objects of class timeSeries using the timeSeries function.

These can be subscripted and used in mathematical operations much like data frames or matrices.

See Also

series class, timeSeries, is.rectangular.

Update Old ts Objects

Description

Converts an old ts object to a signalSeries object.

Usage

ts.update(x)

Arguments

Value

returns a signalSeries object with equivalent positions.

See Also

signalSeries.

Examples

ts.update(ts(1:10)) 

Positions Object Union With Tolerance

Description

Makes a union of numeric or calendar positions (that is, positions of series objects (which can be numeric), time vectors, or sequences) objects using localzone and matchtol as in the seriesMerge and align functions.

Usage

unionPositions(..., localzone = FALSE, matchtol = 0)

Arguments

Value

Returns a new positions object containing all of the input positions, with duplicates (as defined by matchtol and localzone) removed. Returns numeric(0) if no ... arguments are given.

See Also

positions, align, seriesMerge.

Examples

unionPositions(1:10, 5:20)
unionPositions(1:10, 5.1:20.1, matchtol=.3)
unionPositions(timeCalendar(d=1:10), timeCalendar(d=5:20))
unionPositions(timeCalendar(d=1:10, zone="PST"),
               timeCalendar(d=5:20, zone="EST"))
unionPositions(timeCalendar(d=1:10, zone="PST"), 
               timeCalendar(d=5:20, zone="EST"), localzone=TRUE)