Dealing with time series¶
Dealing with radar data typically means implies dealing with time series (of radar records or rain gauge observations). This article gives a brief intro on how to deal with times series and datetimes in Python.
The datetime module¶
The datetime module provides a number of types to deal with dates, times, and time intervals.
import datetime as dt
There are different ways to create datetime objects.
# This is now (system time) now = dt.datetime.now() # Just using the date birth_van_rossum = dt.datetime(1956, 1, 31) # Providing both date and time first_wradlib_commit = dt.datetime(2011, 10, 26, 11, 54, 58) # Or initialising from a string erad_2016_begins = dt.datetime.strptime("2016-10-09 09:00:00", "%Y-%m-%d %H:%M:%S")
You can compute the difference between two datetime objects.
# Age of Guido van Rossum age_van_rossum = now - birth_van_rossum print("This is a %r object.\n" % type(age_van_rossum) ) print("It looks like this: %r" % age_van_rossum ) print("and consists of\n\t%d days,\n\t%d seconds,\n\tand %d microseconds.\n" % (age_van_rossum.days, age_van_rossum.seconds, age_van_rossum.microseconds) ) # Age of wradlib age_wradlib = now - first_wradlib_commit # Time until (or since) beginning of ERAD 2016 OSS Short course from_to_erad2016 = now - erad_2016_begins print("Guido van Rossum is %d seconds old." % age_van_rossum.total_seconds()) print("wradlib's first commit was %d days ago." % age_wradlib.days) if from_to_erad2016.total_seconds() < 0: print("The ERAD 2016 OSS Short course will start in %d days." % -from_to_erad2016.days ) else: print("The ERAD 2016 OSS Short course took place %d days ago." % from_to_erad2016.days)
This is a <class 'datetime.timedelta'> object. It looks like this: datetime.timedelta(days=23675, seconds=47845, microseconds=180524) and consists of 23675 days, 47845 seconds, and 180524 microseconds. Guido van Rossum is 2045567845 seconds old. wradlib's first commit was 3318 days ago. The ERAD 2016 OSS Short course took place 1508 days ago.
Or you can create a
datetime.timedelta object yourself and add/subtract a time interval from/to a
datetime object. You can use any of these keywords:
days, seconds, microseconds, milliseconds, minutes, hours, weeks, but
datetime.timedelta will always represent the result in
days, seconds, microseconds.
# This is an interval of two minutes print(dt.timedelta(minutes=1, seconds=60)) # And this is, too print(dt.timedelta(minutes=2)) now = dt.datetime.now() print("This is now: %s" % now) print("This is two minutes before: %s" % (now - dt.timedelta(minutes=2)) )
0:02:00 0:02:00 This is now: 2020-11-25 13:17:25.196568 This is two minutes before: 2020-11-25 13:15:25.196568
The default string format of a
datetime object corresponds to the isoformat. Using the
strftime function, however, you can control string formatting yourself. The following example shows this feature together with other features we have learned before. The idea is to loop over time and generate corresponding string representations. We also store the
datetime objects in a list.
start = dt.datetime(2016, 10, 9) end = dt.datetime(2016, 10, 14) interval = dt.timedelta(days=1) dtimes =  print("These are the ERAD 2016 conference days (incl. short courses):") while start <= end: print(start.strftime("\t%A, %d. %B %Y")) dtimes.append(start) start += interval
These are the ERAD 2016 conference days (incl. short courses): Sunday, 09. October 2016 Monday, 10. October 2016 Tuesday, 11. October 2016 Wednesday, 12. October 2016 Thursday, 13. October 2016 Friday, 14. October 2016
matplotlib generally understands
datetime objects and tries to make sense of them in plots.
# Instead of %matplotlib inline import matplotlib.pyplot as pl try: get_ipython().magic("matplotlib inline") except: pl.ion() import numpy as np
# Create some dummy data level = np.linspace(100,0,len(dtimes)) # And add a time series plot fig = pl.figure(figsize=(10,5)) ax = fig.add_subplot(111) pl.plot(dtimes, level, "bo", linestyle="dashed") pl.xlabel("Day of the conference", fontsize=15) pl.ylabel("Relative attentiveness (%)", fontsize=15) pl.title("Development of participants' attentiveness during the conference", fontsize=15) pl.tick_params(labelsize=12)