As per Gael's request, here is my progress compared to what was initially stated as mid-term goals.
Overall the project is behind schedule, but not far, and I am fairly confident about its successful completion. -- GOAL: Set up a running performance benchmark such as speed.pypy.org or Wes McKinney’s vbench-generated benchmark for Pandas; this will allow us to publicly make verifiable performance claims. This should run automatically and be up to date with the git master, so that developers can notice when performance regressions occur due to changes to the codebase. STATUS: The scikit-learn-speed project is such a system. The framework in itself is nearly deployable (the configuration is hardcoded and needs to be factored into a config file). It will easily be done by friday. Benchmark module coverage status is around 50%. I aim to cover everything by friday. It's a tough balancing act: scenario coverage vs. benchmark running time, and pruning will be needed. -- GOAL: Efficiently support multiple regression targets (bidimensional Y) in all linear models, like ridge regression and orthogonal matching pursuit currently do. STATUS: Pull request under review. -- -- GOAL: Show the effect of a couple of speedups [...] STATUS: no huge speedup, but a couple of minor ones. See the working notes. https://github.com/scikit-learn/scikit-learn-speed/wiki/Working-notes GOAL: Improve the documentation of all the modules [...] (essentially: summary of how estimators scale) STATUS: POSTPONED, the benchmarking framework needs to be run at least for a couple of days on an independent machine. There is documentation work in progress, specifically addressing what Olivier asked me: > >> Thanks very much for the wrap up Vlad. Could you please document how >> to use the %memit and %mrun tools in the performance chapter of the >> scikit-learn documentation? >> >> http://scikit-learn.org/stable/developers/performance.html > but this doesn't directly answer the stated goal. ------------------ Vlad N. http://vene.ro ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
