Closed kuanb closed 6 years ago
I stupidly went with this pattern when originally implementing MP for the route reducer: https://github.com/kuanb/peartree/blob/eb4fdab7a8484a621a0ee78d7723858b5ca2ac27/peartree/summarizer.py#L269-L301
This loop of appending is what is causing all the performance regression observed! Just request .values each time and extend a reference list, then create a new pandas DataFrame when the whole matrices is constructed.
.values
Such a stupid mistake. Oh, well.
I stupidly went with this pattern when originally implementing MP for the route reducer: https://github.com/kuanb/peartree/blob/eb4fdab7a8484a621a0ee78d7723858b5ca2ac27/peartree/summarizer.py#L269-L301
This loop of appending is what is causing all the performance regression observed! Just request
.values
each time and extend a reference list, then create a new pandas DataFrame when the whole matrices is constructed.Such a stupid mistake. Oh, well.