This would be an immediate fix for the time handling issue when using raw data with different frequencies.
It would need to identify existing time dimensions (%Y and %Y-Q%Q etc.) and enable interpolation between them.
It should also handle inventing a new time dimension--taking yearly and quarterly data and creating monthly, for example.
The output would be a new variable dimensioned by the interpolated date: GDP (%Y) as input becomes GDP_Q (%Y-Q%Q) as output. And so on.
It should support linear and cubic spline interpolation.
The obvious widget would have a jagged line graph on the left and a smooth line graph on the right.
I'm marking this as critical simply because it would impress our current business contact and be very useful for users once we go to sale as well.
A simpler widget still--a zig-zag line and a sine curve on the same mini-plot.
We basically have this widget already. All the binary operations interpolate time values to align to a common grid ("make them conformant") which is a subset of the first argument's xvector. The merge operator does as well, with the subtle effect that the xvector is the union of the two xvectors.
So can I take a yearly series starting in 1834 and a quarterly series
starting in 1946 Andrew ending in 2033 and produce an interpolated
quarterly series from 1834 till 2023?
On Thu, 16 Nov 2023 at 23:40, High Performance Coder hpcoder@users.sourceforge.net wrote:
Ticket moved from /p/minsky/ravel/427/