Date axis labels are off by one

I have a time-series chart that I’d like to specify the tick labels for, but I can’t seem to get the right dates to show up. An MWE looks like this:

import altair as alt
import pandas as pd

axis_dates = ['2020-03-01', '2020-03-08', '2020-03-15', '2020-03-22', '2020-03-29', '2020-04-05']

df = pd.DataFrame({
    'x': axis_dates,
    'y': [1, 3, 5, 6, 4, 2]
})

alt.Chart(df).mark_line().encode(
    x=alt.X('x:T', axis=alt.Axis(title='Date', format=("%b %d"), values=axis_dates)),
    y='y')

How can I make the chart use the dates in the DataFrame? Thank you.

I am using 4.1.0.

EDIT: fixed my code

[jakevdp]

The issue is you are not passing full ISO8601 strings, so some browsers parse them as UTC time, and then they are displayed in local time by Vega-Lite. This is why Altair always outputs full ISO strings (e.g. 2010-01-01T00:00:00) which javascript always parses as local time. In general, the easiest way to deal with this is to always use pandas datetime type for time columns, then Altair takes care of the details for you.

[jakevdp]

e.g.

import altair as alt
import pandas as pd

df = pd.DataFrame({
    'x': pd.to_datetime(['2020-03-01', '2020-03-08', '2020-03-15', '2020-03-22', '2020-03-29', '2020-04-05']),
    'y': [1, 3, 5, 6, 4, 2]
})

alt.Chart(df).mark_line().encode(
    x=alt.X('x:T', axis=alt.Axis(title='Date', format=("%b %d"))),
    y='y'
  )