How to: Plot Time Series

In my activity-tracker project I have simple a list of timestamps. If there is a timestamp, then there is activity.

I would like to create a plot for one day of recorded activity. It should be a a 1440 pixel graphic (24 hours * 60 minutes / hour) x e.g. 20 pixel. Hence x = h * 24 + m (for any y) just indicates if that time is in my timeseries.

Doing it with PIL:

from datetime import datetime, date
from typing import List

def create_day_plot(timestamps: List[datetime], day: date) -> None:
    print(f"{day.strftime('%Y-%m-%d')} -- {len(timestamps)}")
    from PIL import Image
    img = Image.new(mode="RGB", size=(24*60, 20), color='white')

    pixels = img.load() # create the pixel map

    for timestamp in timestamps:
        if timestamp.strftime("%Y-%m-%d") != day.strftime("%Y-%m-%d"):
            continue
        x = timestamp.hour * 60 + timestamp.minute

        for y in range(img.size[1]):
            pixels[x, y] = (255, 0, 0)

    img.save("plot.png")

# Just a tiny sample:
create_day_plot([datetime(2022, 3, 27, 12, 0, 0), datetime(2022, 3, 27, 12, 1, 0), datetime(2022, 3, 27, 12, 2, 0)], date(2022, 3, 27))

With a bit more relaistic data it looks like this:

Is this possible with plotext? (If the answer is “try to make it work with matplotlib first”, then I’ll ask this on StackOverflow, but maybe there is a way to control the output on pixel-level? 😃 )

[piccolomo]

Hi @MartinThoma,

Thanks to your request, I have added the function eventplot(), in the new version 5.1, available only on GitHub for now, using:

pip install git+https://github.com/piccolomo/plotext

The function is described here and you have been credited here.

Any feedback or double test is appreciated 😃.

Thanks and all the best, Savino

[piccolomo]

Hi @MartinThoma,

it occurred to me that you could improve the pixel resolution (double it!) using this other way:

import plotext as plt
from random import randint
from datetime import datetime, timedelta

plt.clf(); 
plt.clt(); # to optionally clean the terminal
plt.datetime.set_datetime_form(date_form="", time_form="%H:%M") # Try also time_form="%H" for clearer x ticks !  

times = [datetime(2022, 3, 27, randint(0, 23), randint(0, 59), randint(0, 59)) for i in range(100)] # A random list of times during the day
times = [plt.datetime.datetime_to_timestamp(el) for el in times] # I will need to fix this; datetime_to_string should have worked.

plt.scatter(times, [1] * len(times), fillx = 1, marker = 'hd', color = 'red')
xticks = [datetime(2022, 3, 27) + timedelta(hours=i) for i in range(25)] 
xticks, xlabels = list(map(plt.datetime.datetime_to_timestamp, xticks)), list(map(plt.datetime.datetime_to_string, xticks))
plt.xticks(xticks, xlabels); plt.yticks();
plt.xlim(xticks[0], xticks[-1]); plt.ylim(0, 1);
plt.frame(0)
plt.xaxis(1, "lower"); plt.xaxis(1, "upper")
plt.xlabel("Hours in The Day")
plt.title("Activity Tracker")
plt.plotsize(None, 20) # Set the height you prefer or comment for maximum size 
plt.show()

with this result: