time-series-average.py

#!/usr/bin/env python3

import sys

if '-h' in sys.argv or '--help' in sys.argv or len(sys.argv) < 2:
    print('''
Usage: {} run
(the run keyword just indicates you read/know this)

This script calculates what, as far as I've been able to come up with and find,
is hopefully the best possible average value during a time series when the 
times are neither regular nor random. In either case, you could just average
the values, but if there is bias (like if you check your freezer more often
after you noticed the temperature is off), this script should correct for that
bias. The result will not be correct, but it's hopefully the least wrong in
the given scenario.

Expects stdin formatted as: unixtime,value\\nunixtime,value\\n (etc.)

Example:
 0,0
10,5
11,99

value
 0                         5  99
||-----------||-------------||-|
 1  2  3  4  5  6  7  8  9 10 11
time

Will result in 7 because time 0-5 is considered to be of value zero, time
5-10.5 is considered to be of value 5, and 10.5-11 is considered to be of value
99. Over 11 seconds, that total comes to 5*0 + 5*5.5 + 99*0.5 = 77, divided by 11 is
an average of 7.
'''.lstrip().format(sys.argv[0]))
    exit(1)

first = True
total = 0
starttime = None
for line in sys.stdin:
    if ',' not in line:  # blank line, maybe a comment, idk
        continue

    t, val = map(float, line.strip().split(','))
    if first:
        starttime = t
        lasttime = t
        lastvalue = val

        first = False
        continue

    timepassed = t - lasttime
    total += lastvalue * (timepassed / 2)
    total += val * (timepassed / 2)

    lasttime = t
    lastvalue = val

print(total / (lasttime - starttime))