Description
We are in the process of upgrading PVLib from v0.5.2 to v0.6.3, and caught some big changes in the DIRINT and DIRINDEX functions. Here are some plots ran from the same notebook with both versions:
0.5.2
0.6.3
The GHI clearsky (used by the DIRINDEX model) seems to be kind of shifted compared to 0.5.2:
DIRINT changed compared to 0.5.2 and DIRINDEX is way off:
Is this expected? Or are we doing anything wrong? It seems it is due to the changes in #400. Wanted to share this before digging further into the root cause.
Here is the notebook, and below are the data.