Re: [Numpy-discussion] 2d binning and linear regression

2010-06-22 Thread Tom Durrant
> > > > > What exactly are trying to fit because it is rather bad practice to fit > a model to some summarized data as you lose the uncertainty in the > original data? > If you define your boxes, you can loop through directly on each box and > even fit the equation: > > model=mu +beta1*obs > > The

Re: [Numpy-discussion] 2d binning and linear regression

2010-06-22 Thread Tom Durrant
> > > the basic idea is in "polyfit on multiple data points" on > numpy-disscusion mailing list April 2009 > > In this case, calculations have to be done by groups > > subtract mean (this needs to be replaced by group demeaning) > modeldm = model - model.mean() > obsdm = obs - obs.mean() > > xx =

Re: [Numpy-discussion] 2d binning and linear regression

2010-06-20 Thread Tom Durrant
> > are you doing something like np.polyfit(model, obs, 1) ? > > If you are using polyfit with deg=1, i.e. fitting a straight line, > then this could be also calculated using the weights in histogram2d. > > histogram2d (histogramdd) uses np.digitize and np.bincount, so I'm > surprised if the hi

[Numpy-discussion] 2d binning and linear regression

2010-06-20 Thread Tom Durrant
Hi All, I have a problem involving lat/lon data. Basically, I am evaluating numerical weather model data against satellite data, and trying to produce gridded plots of various statistics. There are various steps involved with this, but basically, I get to the point where I have four arrays of th