limit how much data is read with numpy.genfromtxt for matplotlib

matplotlib, numpy, python

Solution

`numpy.genfromtxt` accepts iterators as well as files. That means it will accept the output of `itertools.islice`. Here, `test.txt` is a five-line file:

>>> import itertools, numpy
>>> with open('test.txt') as t_in:
...     numpy.genfromtxt(itertools.islice(t_in, 3))
... 
array([[  1.,   2.,   3.,   4.,   5.],
       [  6.,   7.,   8.,   9.,  10.],
       [ 11.,  12.,  13.,  14.,  15.]])

One might think this would be slower than letting `numpy` handle the file IO, but a quick test suggests otherwise. `genfromtxt` provides a `skip_footer` keyword argument that you can use if you know how long the file is...

>>> numpy.genfromtxt('test.txt', skip_footer=2)
array([[  1.,   2.,   3.,   4.,   5.],
       [  6.,   7.,   8.,   9.,  10.],
       [ 11.,  12.,  13.,  14.,  15.]])

...but a few informal tests on a 1000-line file suggest that using `islice` is faster even if you skip only a few lines:

>>> def get(nlines, islice=itertools.islice):
...     with open('test.txt') as t_in:
...         numpy.genfromtxt(islice(t_in, nlines))
...         
>>> %timeit get(3)
1000 loops, best of 3: 338 us per loop
>>> %timeit numpy.genfromtxt('test.txt', skip_footer=997)
100 loops, best of 3: 4.92 ms per loop
>>> %timeit get(300)
100 loops, best of 3: 5.04 ms per loop
>>> %timeit numpy.genfromtxt('test.txt', skip_footer=700)
100 loops, best of 3: 8.48 ms per loop
>>> %timeit get(999)
100 loops, best of 3: 16.2 ms per loop
>>> %timeit numpy.genfromtxt('test.txt', skip_footer=1)
100 loops, best of 3: 16.7 ms per loop

Problem

I am creating a graph in python using a text file for the source data and matplotlib to plot the graph. The simple logic below works well. But is there a way to get have numpy.gentfromtxt only read the first 50 lines from the file 'temperature_logging'? Currently it reads the entire file. ``` temp = numpy.genfromtxt('temperature_logging',dtype=None,usecols=(0)) time = numpy.genfromtxt('temperature_logging',dtype=None,usecols=(1)) dates = matplotlib.dates.datestr2num(time) pylab.plot_date(dates,temp,xdate=True,fmt='b-') pylab.savefig('gp.png') ``` contents in temperature_logging; ``` 21.75 12-01-2012-15:53:35 21.75 12-01-2012-15:54:35 21.75 12-01-2012-15:55:35 . . . ```

Original source