bug: IndicatorFactory - Unnecessary column level while operating with single column

[andreas-vester]

I am making use of the sample from the documentation (https://vectorbt.dev/api/indicators/factory/#naive-approach), with the exception that my price only consists of one security (column).

from datetime import datetime

import pandas as pd
import vectorbt as vbt

price = pd.DataFrame(
    {"a": [1, 2, 3, 4, 5]},
    index=pd.Index(
        [
            datetime(2020, 1, 1),
            datetime(2020, 1, 2),
            datetime(2020, 1, 3),
            datetime(2020, 1, 4),
            datetime(2020, 1, 5),
        ]
    ),
).astype(float)

MyInd = vbt.IndicatorFactory(
    input_names=["price"],
    param_names=["window"],
    output_names=["ma"],
).from_apply_func(vbt.nb.rolling_mean_nb)

When running the indicator with any kind of window combination, the resulting dataframe will consist of multiple columns. Everything is fine here.

myind1.ma

custom_window    2    3
2020-01-01     NaN  NaN
2020-01-02     1.5  NaN
2020-01-03     2.5  2.0
2020-01-04     3.5  3.0
2020-01-05     4.5  4.0

As soon as I reduce the number of parameters to a single parameter, the output looks a bit different, i.e. there has been added another column level with no name and the value a.

myind2.ma

custom_window    2
                 a
2020-01-01     NaN
2020-01-02     1.5
2020-01-03     2.5
2020-01-04     3.5
2020-01-05     4.5

This might causing issues when further processing these results. I wouldn't expect any difference in the column levels no matter how many columns.

Is there an easy way to avoid this behaviour?

[polakowo]

You could set drop_redundant=False in vbt.settings.broadcasting to keep "a" in both examples.

[andreas-vester]

I would rather like to get rid of "a" in both examples. Is this possible?

[github-actions]

This issue is stale because it has been open for 90 days with no activity.