arima.fit performance #713

Closed
jseabold opened this Issue Mar 20, 2013 · 4 comments

Projects

None yet

1 participant

@jseabold
Member

It looks like predict() maybe be called at each step of the iteration in ARIMA with all the dates making machinery too. Need a dates-free implementation for "fittedvalues" or something for the estimation to be quicker.

@jseabold
Member

Correction. It's in the determination of start_params not in the actual fitting process.

@jseabold
Member

Parking this here. The (out of sample) forecasting functions are a good candidate for Cython.

For the Kalman Filter, we should explore two speed-ups.

  1. using memoryviews for accessing the underlying arrays without python overhead

http://jakevdp.github.com/blog/2012/08/08/memoryview-benchmarks/
http://jakevdp.github.com/blog/2012/08/16/memoryview-benchmarks-2/

  1. and/or calling lapack directly without accessing python. I'm not positive I did it this way, passing pointers, the last time I tried this.

http://matforge.org/fipy/browser/fipy/fipy/tools/smallMatrixVectorOpsExt.pyx?rev=3ba0cc448b1c6

@jseabold
Member
jseabold commented Sep 4, 2013

See #1069.

@jseabold
Member
jseabold commented Apr 6, 2014

Closed by a738b4f. There's another open issue for start_params so I'll look into that when I get to it. See #1301 and #1366.

@jseabold jseabold closed this Apr 6, 2014
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment