How can I convert from YYYY-MM-DD on datetime64 to MM-DD and plotting tips

Question

I have a database from COVID that I'm playing with. I have info in the format YYYY-MM-DD and I'm plotting days vs. number of cases in a bar plot. In my graph, the days appear as months, but I want an axis with every 7 days. Should I convert my datetime64 format to MM-DD and plot my graph?

This is the code I'm using to plot:

fig, ax = plt.subplots(figsize=(15,7))
ax.bar(x0, y0, color='green')

ax.set(xlabel = 'Data da notificação', 
       ylabel = 'Casos novos')

plt.setp(ax.get_xticklabels(), rotation = 45)

plt.show()

And this is the resulted graph:

This is the graph I want:

Sample data: https://docs.google.com/spreadsheets/d/1q4Njy0mGMtIeEABdaFrzNEMSavjpXtlUn8BM8z_QZBk/edit?usp=sharing

Can you add some sample data so we can reproduce the graph? – marsnebulasoup Aug 21 '20 at 00:20 — marsnebulasoup, Aug 21 '20 at 00:20
Done! It is at the end of the question – Eduardo Valentin dos Santos Aug 21 '20 at 00:34 — Eduardo Valentin dos Santos, Aug 21 '20 at 00:34
I would also like to fix these weird limits of the graph – Eduardo Valentin dos Santos Aug 21 '20 at 03:05 — Eduardo Valentin dos Santos, Aug 21 '20 at 03:05

Ynjxsjmh · Accepted Answer · 2020-08-21T15:08:42.503

To format like MM-DD, you can specify fmt argument which accepts a strftime format string for formatting and is a required argument to the DateFormatter.

You can also specify an interval argument to the DateLocator. e.g. interval=7 the locator places ticks at every 7th date.

import datetime
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.dates as mdates


df = pd.read_csv("1.csv", delim_whitespace=True)

df['data'] = pd.to_datetime(df['data'], format='%m/%d/%Y')

fig, ax = plt.subplots(figsize=(10, 6))

ax.bar(df['data'], df['casosNovos'])

ax.xaxis.set_major_locator(mdates.DayLocator(interval=7))
ax.xaxis.set_major_formatter(mdates.DateFormatter(fmt='%m%d'))

ax.set_xlim(df['data'].min(), datetime.date(2020, 8, 21))

plt.setp(ax.get_xticklabels(), rotation = 45)

plt.show()

1.csv looks like

regiao  data    casosNovos
Brasil  2/25/2020   0
Brasil  2/26/2020   1
Brasil  2/27/2020   0
Brasil  2/28/2020   0
Brasil  2/29/2020   1
Brasil  3/1/2020    0
Brasil  3/2/2020    0
Brasil  3/3/2020    0

Cool! How do I limit the x axis in datetime? The axis extends to dates that do not exist in the database. — Eduardo Valentin dos Santos, Aug 21 '20 at 14:25
@EduardoValentindosSantos Use `ax.set_xlim()`. See my updated answer. — Ynjxsjmh, Aug 21 '20 at 15:09

How can I convert from YYYY-MM-DD on datetime64 to MM-DD and plotting tips

1 Answers1