How can one plot time-related information with python in a natural human-readable way?

Question

Often, you have a set of events (visits to a website, price information, whatever) that's related to a time value, let's say a timestamp (though a datetime object is just fine). How can they be plotted so that the time axis gets human-readable, meaningful values instead of just number of seconds?

I've been looking around gnuplot and matplot but I haven't been able to find quite the way to do this. The problem is, while it's possible for matplot to set ticks every hour, it would be better to be able to see textual time info every N hours instead of having to count them up.

I suspect gnuplot is overkill/not really designed for this. Any suggestions?

Answer 1

FWIW here's a simple example of plotting some values against times using the excellent and well-documented matplotlib:

data.csv:
    VISIT_TIME  TOTAL_VISITS
    06:00:00    290
    06:30:00    306
    07:00:00    364
    07:30:00    363
    08:00:00    469
    08:30:00    436
    09:00:00    449
    09:30:00    451
    10:00:00    524
    10:30:00    506
    11:00:00    613
    11:30:00    585
    12:00:00    620
    12:30:00    529
    13:00:00    588
    13:30:00    545

Simple program for illustrative purposes:

import matplotlib.dates as mdates
import matplotlib.mlab as mlab
import matplotlib.pyplot as plt
import datetime as dt
import sys

def main( datafile ):
    np_rec_array = mlab.csv2rec( datafile, delimiter='\t' )
    np_rec_array.sort() # in-place sort
    # a `figure` is a starting point for MPL visualizations
    fig = plt.figure( figsize=(8,6) ) 
    # add a set of `axes` to above `figure`
    ax = fig.add_subplot(111)
    x = np_rec_array.visit_time
    y = np_rec_array.total_visits
    # `plot_date` is like `plot` but allows for easier x-axis formatting
    ax.plot_date(x, y, 'o-', color='g') 
    # show time every 30 minutes
    ax.xaxis.set_major_locator( mdates.MinuteLocator(interval=30) )
    # specify time format
    ax.xaxis.set_major_formatter( mdates.DateFormatter("%H:%M") )
    # set x-axis label rotation (otherwise they can overlap)
    for l in ax.get_xticklabels():
        l.set_rotation(60)
    plt.title( 'Website Visits' )
    plt.show()

if __name__ == '__main__':
    if len( sys.argv ) == 1:
        sys.stderr.write( 'need a filename, exiting...' )
        sys.exit(-1)
    main( sys.argv[1] )

Output is the following image: