如何仅绘制熊猫 datetime64[ns] 属性的时间

Question

I have a dataframe of a long time range in format datetime64[ns] and a int value

Data looks like this:

                  MIN_DEP  DELAY
0     2018-01-01 05:09:00      0
1     2018-01-01 05:13:00      0
2     2018-01-01 05:39:00      0
3     2018-01-01 05:43:00      0
4     2018-01-01 06:12:00     34
...                   ...    ...
77005 2020-09-30 23:42:00      0
77006 2020-09-30 23:43:00      0
77007 2020-09-30 23:43:00     43
77008 2020-10-01 00:18:00      0
77009 2020-10-01 00:59:00      0

[77010 rows x 2 columns]
MIN_DEP    datetime64[ns]
DELAY               int64
dtype: object

Target is to plot all the data in just a 00:00 - 24:00 range on the x-axis, no dates anymore.

As i try to plot it, the timeline is 00:00 at any point. How to fix this?

import matplotlib.dates as mdates

fig, ax = plt.subplots()
ax.plot(pd_to_stat['MIN_DEP'],pd_to_stat['DELAY'])

xfmt = mdates.DateFormatter('%H:%M')
ax.xaxis.set_major_formatter(xfmt)

plt.show()

tried to convert the timestamps before to dt.time and plot it then

pd_to_stat['time'] = pd.to_datetime(pd_to_stat['MIN_DEP'], format='%H:%M').dt.time
fig, ax = plt.subplots()
ax.plot(pd_to_stat['time'],pd_to_stat['DELAY'])
plt.show()

Plot does not allow to do that:

TypeError: float() argument must be a string or a number, not 'datetime.time'

Answer 1

According to your requirement, I guess you don't need the dates and as well as the seconds field in your timestamp. So you need a little bit of preprocessing at first. Remove the seconds field using the code below

dataset['MIN_DEP'] = dataset['MIN_DEP'].strftime("%H:%M")

Then you can remove the date from your timestamp in the following manner

dataset['MIN_DEP'] = pd.Series([val.time() for val in dataset['MIN_DEP']])

Then you can plot your data in the usual manner.

Answer 2

This seems to work now. I did not recognise, the plot was still splitting up in dates. To work around I hat to replace all the dates with the same date and plottet it hiding the date using DateFormatter

import matplotlib.dates as mdates
pd_to_stat['MIN_DEP'] = pd_to_stat['MIN_DEP'].map(lambda t: t.replace(year=2020, month=1, day=1))

fig, ax = plt.subplots()
ax.plot(pd_to_stat['MIN_DEP'],pd_to_stat['DELAY'])

xfmt = mdates.DateFormatter('%H:%M')
ax.xaxis.set_major_formatter(xfmt)

plt.show()