Suppose I have looped through N time series variables and have N slope
coefficients from a simple regression for each dependent variable in a Nx1
matrix.
I now want to show these as y variables in a scatter plot for a given x
variable -- in a different (and already constructed) Nx1 matrix. Can I
easily do this ?
I tricked GRETL to plot 2 new series based on the matrices as so
scalar nn=rows(bx)
smpl 2001:01 2001:01+nn-1
series y= bx
series x= mx
chart1 <- gnuplot y x --with-lines
But I'd rather not trick gretl by creating series, when the matrix values
are not actually time series.
I could export the matrices to a new dataset or export to another plotting
package or even use EXCEL, but I wonder if there is something more elegant
all in GRETL
-- i.e. create a plot directly from the matrices, independent of the
dataset design (ie. time series, cross section, panel) and current smpl
period
PS It took me some time to figure out I could only create the series y=
bx if the smpl was the exact right size -- error message was a little
obtuse. The error: 'The variable y is of type series'
A error message such as 'Can not create series from matrix unless smpl
period agrees with matrix size' would be bettter.