psmat function

Matrix / Array from Panel Series

psmat efficiently expands a panel-vector or 'indexed_series' ('pseries') into a matrix. If a data frame or 'indexed_frame' ('pdata.frame') is passed, psmat returns a 3D array or a list of matrices.

psmat(x, ...)

## Default S3 method:
psmat(x, g, t = NULL, transpose = FALSE, fill = NULL, ...)

## S3 method for class 'data.frame'
psmat(x, by, t = NULL, cols = NULL, transpose = FALSE, fill = NULL, array = TRUE, ...)

# Methods for indexed data / compatibility with plm:

## S3 method for class 'pseries'
psmat(x, transpose = FALSE, fill = NULL, drop.index.levels = "none", ...)

## S3 method for class 'pdata.frame'
psmat(x, cols = NULL, transpose = FALSE, fill = NULL, array = TRUE,
      drop.index.levels = "none", ...)

## S3 method for class 'psmat'
plot(x, legend = FALSE, colours = legend, labs = NULL, grid = FALSE, ...)

Arguments

• x: a vector, indexed series 'indexed_series' ('pseries'), data frame or 'indexed_frame' ('pdata.frame').
• g: a factor, GRP object, atomic vector (internally converted to factor) or a list of vectors / factors (internally converted to a GRP object) used to group x. If the panel is balanced an integer indicating the number of groups can also be supplied. See Examples.
• by: data.frame method: Same input as g, but also allows one- or two-sided formulas using the variables in x, i.e. ~ idvar or var1 + var2 ~ idvar1 + idvar2.
• t: same inputs as g/by, to indicate the time-variable(s) or second identifier(s). g and t together should fully identify the panel. If t = NULL, the data is assumed sorted and seq_col is used to generate rownames for the output matrix.
• cols: data.frame method: Select columns using a function, column names, indices or a logical vector. Note: cols is ignored if a two-sided formula is passed to by.
• transpose: logical. TRUE generates the matrix such that g/by -> columns, t -> rows. Default is g/by -> rows, t -> columns.
• fill: element to fill empty slots of matrix / array if panel is unbalanced. NULL will generate a NA of the right type.
• array: data.frame / pdata.frame methods: logical. TRUE returns a 3D array (if just one column is selected a matrix is returned). FALSE returns a list of matrices.
• drop.index.levels: character. Either "id", "time", "all" or "none". See indexing .
• ...: arguments to be passed to or from other methods, or for the plot method additional arguments passed to ts.plot.
• legend: logical. Automatically create a legend of panel-groups.
• colours: either TRUE to automatically colour by panel-groups using rainbow or a character vector of colours matching the number of panel-groups (series).
• labs: character. Provide a character-vector of variable labels / series titles when plotting an array.
• grid: logical. Calls grid to draw gridlines on the plot.

Details

If n > 2 index variables are attached to an indexed series or frame, the first n-1 variables in the index are interacted.

Returns

A matrix or 3D array containing the data in x, where by default the rows constitute the groups-ids (g/by) and the columns the time variable or individual ids (t). 3D arrays contain the variables in the 3rd dimension. The objects have a class 'psmat', and also a 'transpose' attribute indicating whether transpose = TRUE.

Note

The pdata.frame method only works for properly subsetted objects of class 'pdata.frame'. A list of 'pseries' won't work. There also exist simple aperm and [ (subset) methods for 'psmat' objects. These differ from the default methods only by keeping the class and the 'transpose' attribute.

See Also

Time Series and Panel Series , Collapse Overview

Examples

## World Development Panel Data
head(wlddev)                                                    # View data
qsu(wlddev, pid = ~ iso3c, cols = 9:12, vlabels = TRUE)         # Sumarizing data
str(psmat(wlddev$PCGDP, wlddev$iso3c, wlddev$year))             # Generating matrix of GDP
r <- psmat(wlddev, PCGDP ~ iso3c, ~ year)                       # Same thing using data.frame method
plot(r, main = vlabels(wlddev)[9], xlab = "Year")               # Plot the matrix
str(r)                                                          # See srructure
str(psmat(wlddev$PCGDP, wlddev$iso3c))                          # The Data is sorted, could omit t
str(psmat(wlddev$PCGDP, 216))                                   # This panel is also balanced, so
# ..indicating the number of groups would be sufficient to obtain a matrix

ar <- psmat(wlddev, ~ iso3c, ~ year, 9:12)                      # Get array of transposed matrices
str(ar)
plot(ar)
plot(ar, legend = TRUE)
plot(psmat(collap(wlddev, ~region+year, cols = 9:12),           # More legible and fancy plot
           ~region, ~year), legend = TRUE,
     labs = vlabels(wlddev)[9:12])

psml <- psmat(wlddev, ~ iso3c, ~ year, 9:12, array = FALSE)     # This gives list of ps-matrices
head(unlist2d(psml, "Variable", "Country", id.factor = TRUE),2) # Using unlist2d, can generate DF

## Indexing simplifies things
wldi <- findex_by(wlddev, iso3c, year)  # Creating an indexed frame
PCGDP <- wldi$PCGDP                     # An indexed_series of GDP per Capita
head(psmat(PCGDP), 2)                   # Same as above, more parsimonious
plot(psmat(PCGDP))
plot(psmat(wldi[9:12]))
plot(psmat(G(wldi[9:12])))              # Here plotting panel-growth rates