Cannot Animate Plot by pandas.Timestamp

[johnmccain]

Attempting to animate a plot using a pandas Timestamp column results in an exception.

Using Python 3.6.8, plotly_express==0.4.1, plotly==4.1.0

Code to replicate:

import plotly_express as px
import pandas as pd

# create a dataframe with mock data
df = pd.DataFrame(
    [
        {"x": 1, "y": 1, "date": "2018-01-01"},
        {"x": 2, "y": 1, "date": "2018-01-02"},
        {"x": 3, "y": 1, "date": "2018-01-03"},
    ]
)
df["date"] = pd.to_datetime(df["date"])
df.head()

#       date | x | y
# 2018-01-01 | 1 | 1
# 2018-01-02 | 2 | 1
# 2018-01-03 | 3 | 1

# attempt to plot
px.scatter(df, x="x", y="y", animation_frame="date")

Exception & stack trace:

Traceback (most recent call last):
  File "<stdin>", line 2, in <module>
  File "C:\Users\jmccain\AppData\Local\Programs\Python\Python36\lib\site-packages\plotly\express\_chart_types.py", line 52, in scatter
    return make_figure(args=locals(), constructor=go.Scatter)
  File "C:\Users\jmccain\AppData\Local\Programs\Python\Python36\lib\site-packages\plotly\express\_core.py", line 874, in make_figure
    orders, sorted_group_names = get_orderings(args, grouper, grouped)
  File "C:\Users\jmccain\AppData\Local\Programs\Python\Python36\lib\site-packages\plotly\express\_core.py", line 859, in get_orderings
    key=lambda g: orders[col].index(g[i]) if g[i] in orders[col] else -1,
  File "C:\Users\jmccain\AppData\Local\Programs\Python\Python36\lib\site-packages\plotly\express\_core.py", line 859, in <lambda>
    key=lambda g: orders[col].index(g[i]) if g[i] in orders[col] else -1,
ValueError: Timestamp('2018-01-01 00:00:00') is not in list

I found the cause of the issue, in _core.py:847

Pandas .unique() method returns a NumPy array, and as given in the examples section, it will convert a pandas.Timestamp to a numpy.datetime64.

>>> pd.Series([pd.Timestamp('2016-01-01') for _ in range(3)]).unique()
array(['2016-01-01T00:00:00.000000000'], dtype='datetime64[ns]')

The uniques list defined at _core.py:847 uses the .unique() method, converting the series of pandas.Timestamp into an array of numpy.datetime64

uniques = args["data_frame"][col].unique()

This distinction is relevant on line 857:

group_names = sorted(
    group_names,
    key=lambda g: orders[col].index(g[i]) if g[i] in orders[col] else -1
)

Which causes the ValueError when .index() does not find a matching element due to this behavior of pandas.Timestamp and numpy.datetime64 seen here:

>>> pandas.Timestamp('2018-01-01 00:00:00') in [numpy.datetime64('2018-01-01T00:00:00.000000000')]
True
>>> [numpy.datetime64('2018-01-01T00:00:00.000000000')].index(pandas.Timestamp('2018-01-01 00:00:00'))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: Timestamp('2018-01-01 00:00:00') is not in list

[johnmccain]

Having dug a little deeper, the issue around animating by pandas.Timestamp goes beyond the code segment I addressed above. Is animating by date supported functionality? If not, would this be better as a feature request?

[toships]

I have the same issue

workaround that works for me:
df_summary_time = df_summary_time.sort_values(by='timestamp') # sort rows based on timestamp

df_summary_time['timestamp'] = df_summary_time.timestamp.apply(lambda x: x.date()).apply(str) # convert timestamp to a string

Then in px.scatter use animation_frame="timestamp"

[javid-jouzdani]

Thank you toships,
worked for me too.

[robroc]

I have this problem too. A simpler workaround that worked for me is passing:
animation_frame = df.datecolumn.astype(str)

[JorgeMiguelGomes]

I have this problem too. A simpler workaround that worked for me is passing:
animation_frame = df.datecolumn.astype(str)

This works perfectly! Thank you!

[riyadparvez]

Is there any plan to fix this issue?

[gvwilson]

Hi - we are currently trying to tidy up Plotly's public repositories to help us focus our efforts on things that will help users most. Since this issue has been sitting for several years, I'm going to close it. If it's still a concern, we'd be grateful if you could open a new issue (with a short reproducible example if appropriate) so that we can add it to our backlog. Thanks for your help - @gvwilson