search

microprediction / timemachines

Predict time-series with one line of code.
https://www.microprediction.com/blog/popular-timeseries-packages
MIT License
398 stars 51 forks source link

alloptimizers.py :add command line args, collect results into table #7

Closed druce closed 3 years ago

druce commented 3 years ago

took a crack at adding some command line args for alloptimizers.py cleaned up output a little collect results into a logfile CSV

on further review probably just as well to let individual algos be multithreaded or not

no worries if this is not the preferred way but maybe this helps as a start

usage: alloptimizers.py [-h] [-d NDIMS [NDIMS ...]] [-t NTRIALS [NTRIALS ...]] [-v] [-o LOGFILE]

Run all optimizers on input size ndim (default 2) requesting ntrials (default 20) iterations and save results (default log.csv

optional arguments: -h, --help show this help message and exit -d NDIMS [NDIMS ...], --ndims NDIMS [NDIMS ...] Number of input dimensions to objective function (default 2) -t NTRIALS [NTRIALS ...], --ntrials NTRIALS [NTRIALS ...] Number of trial iterations in optimization (default 20) -v, --verbose Increase output verbosity -o LOGFILE, --logfile LOGFILE Specify outputfile (default log.csv)

microprediction commented 3 years ago

You'll like this too. https://github.com/microprediction/timemachines-testing/tree/main/optimizer_elo_ratings/leaderboards/overall