A command-line tool for creating plots from data in text files.
pip install lmj.plot
Or, clone this repository and put the plot script somewhere in your
git clone http://github.com/lmjohns3/py-grep-plot export PATH=$PATH:$(pwd)/py-grep-plot/scripts
Let's say you're running an experimental algorithm, and you put accuracy values in a log file as the experiments run. Here's a snippet from an example log file:
D 2012-03-19 15:02:35,181 decoded p-a-n-c-r-e-a-t-i-c in 4058ms D 2012-03-19 15:02:35,365 tags p-ae2-n-k-r-iy0-ae1-t-ih0-k, best p-ae1-n-k-er0-_-eh1-th-iy0-_ D 2012-03-19 15:02:35,591 averaged 22932 weights in 786ms D 2012-03-19 15:02:35,802 decoded g-y-r in 998ms D 2012-03-19 15:02:36,054 tags jh-ay1-r, best g-_-er0 I 2012-03-19 15:02:36,055 training accuracy: 39.63 D 2012-03-19 15:02:36,246 averaged 23056 weights in 643ms D 2012-03-19 15:02:36,295 decoded s-p-i-t-z-l-e-y in 4090ms D 2012-03-19 15:02:36,540 tags s-p-ih1-t-s-l-_-iy0, best s-p-ey1-t-ah0-l-_-iy0
All of those "training accuracy" lines hidden in there will give us a good idea of how well the algorithm is performing. To get a quick plot of them:
cat ~/Experiments/tagger-beam1.log | py-grep-plot 'training accuracy: ([.\d]+)'
If you have your matplotlib configured with an interactive backend, you should see a nice little plot appear.
The general usage of the script is
py-grep-plot [regexp] < file
Basically, you provide a bunch of data on stdin, and a regular expression that specifies how to extract data from the files. The plotting script will check the regular expression against each input line, parsing out numerical values from those that match. Each matched value will be included in the plot.
If you just provide one match group in your regular expression, the matched values will be plotted on the ordinate, in data-file order. If you want explicit control over the abscissa, just include another match group in your regular expression:
nl ~/Experiments/tagger-beam1.log | py-grep-plot '^(\d+) .* training accuracy: ([.\d]+)'
nl utility numbers the lines of the input file.)
If you provide three match groups per line, the first is plotted along the abscissa, the second along the ordinate, and the third gives the size of an error bar along the ordinate.
You can also provide multiple input files, and the script will show multiple data series on the same plot:
py-grep-plot [regexp] [file]...
Each file will use the same regular expression for matching data.
You can smooth the ordinates by using either the
-s N (
--smooth N) or the
-b N (
--batch N) options. The
--smooth option convolves a rectangular
filter over the data values before plotting, which yields smoother curves but
has edge effects. The
--batch option groups the input data and plots just the
mean and standard deviation of each group.
There are several other command-line options, including control over the plot
colors and styles, X- and Y-axis limits, ; use
--help to get an overview.
(The MIT License)
Copyright (c) 2011-2012 Leif Johnson firstname.lastname@example.org
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the 'Software'), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED 'AS IS', WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.