Pluggable query_strategies

LSSTDESC / RESSPECT

The RESSPECT project is a result from an inter-collaboration agreement established between the Cosmostatistics Initiative (COIN) and the LSST Dark Energy Science Collaboration (DESC) with the goal of developing a recommendation system for telescope resource allocation able to optimize photometric supernova cosmology anaylsis.

MIT License

1 stars 0 forks source link

Pluggable query_strategies #59

Closed drewoldag closed 1 month ago

drewoldag commented 1 month ago

This PR represents the work in implement pluggable query strategies. It is very similar is style to the pluggable classifier work.

A QueryStrategy class has been created
Subclasses of QueryStrategy have been created for each of the 7 different built in strategies.
Machinery is in place to support and use query strategies that are defined in an external library.
Updated tests

github-actions[bot] commented 1 month ago

Before [71af541d]	After [1eeb3a51]	Ratio	Benchmark (Parameter)
158±2ms	159±2ms	1.01	benchmarks.time_feature_creation
174±3ms	175±6ms	1.01	benchmarks.time_learn_loop('KNN', 'UncSampling')
196M	196M	1	benchmarks.peakmem_learn_loop('KNN')
188M	188M	1	benchmarks.peakmem_learn_loop('RandomForest')
171±3ms	172±2ms	1	benchmarks.time_learn_loop('KNN', 'RandomSampling')
2.65±0.01s	2.65±0.02s	1	benchmarks.time_learn_loop('RandomForest', 'RandomSampling')
2.68±0.02s	2.69±0.03s	1	benchmarks.time_learn_loop('RandomForest', 'UncSampling')

Click here to view all benchmarks.