DataFrame with DatetimeIndex is shown as RangeIndex

[juanfcocontreras]

When I try to show a DataFrame that has an index of type DatetimeIndex, in dtale it is shown as a RangeIndex.

It would be very useful to display that DatetimeIndex.

Sample code:

import pandas as pd
import pandas._testing as tm

df = tm.makeTimeDataFrame()

df

from dtale import show

show(df)

[aschonfeld]

So in order to get this to work correctly you would have to updated your code to this:

import dtale
import pandas._testing as tm

df = tm.makeTimeDataFrame()
df.index.name = 'date_index'
dtale.show(df)

The problem is that on startup it doesn't maintain indexes because it would be too hard to keep track of. So it runs .reset_index() on any startup and drops any columns names index by default. Because in most instances users are giving dataframes with index names and it will the resulting columns after the reset_index() call will be pre-locked on the grid (to show the users they were the original index).

The best option I can offer other than the code fix above is that I can check to see if the index on the dataframe is not dtype('int64') (which is the default 0 to n-1 natural index in pandas) and if it isn't then I will allow the default naming convention of "index" to pass through for those.

But the best solution is to just include a name on any index column that isn't "index".

Hope this helps!

[juanfcocontreras]

The best option I can offer other than the code fix above is that I can check to see if the index on the dataframe is not dtype('int64') (which is the default 0 to n-1 natural index in pandas) and if it isn't then I will allow the default naming convention of "index" to pass through for those.

I think that alternative would be great, but while this workaround is enough for me.

Thank you very much!

[juanfcocontreras]

I think it would be better to do the following check instead of df.index.dtype != "int64". This way you could have a custom RangeIndex and it would be kept.

if not df.index.equals(pd.RangeIndex(0, len(df))):
...

[aschonfeld]

Added v1.8.12