Set values for time axis

I’m unclear as to how I might set the values for a time axis. I’m using Axis as follows…

Axis(title=my_title, values=my_values)

…and have tried setting my_values to a list of ISO formatted strings, datetime objects, datetime64 objects (np) and a DatetimeIndex object (pd). In each case, the stacktrace indicates that a float is required.

Thanks for any help in advance. I appreciate that this is less an issue and more a request for assistance, but there isn’t an Altair discussion forum.

[kanitw]

This reminds me that we should support our DateTime Definition object (that we introduced for temporal filter) for axis/legend’s values as well (https://github.com/vega/vega-lite/issues/1516). So this will be reasonable to use when people use Vega-Lite from JSON.

That said, I think from Altair’s perspective, I think automatically date conversion to timestamp would be better. 😃

probably oweing to differing treatments of midnight between pandas and vega-lite.

Possibly because the timestamp is UTC but our label displays local time.
We’ll have to work on support for UTC (https://github.com/vega/vega-lite/issues/446).

cc: @domoritz

[iaindillingham]

Hello everyone,

Thank you all for your time. It’s reassuring to find such a committed bunch of core developers!

I followed @jakevdp and passed timestamps to Axis. Rather than hard-coding dates, I used pd.date_range as follows:

date_min = '2016-01-01'
date_max = '2016-02-01'
my_values = list(pd.date_range(date_min, date_max, freq='W-MON').astype(int) / 10 ** 6)

The curious can find the associated notebook here. (It’s a quick exploration of bike usage in Bath, UK, which I prepared for an upcoming hack day.)