[BUG] inconsistent date/time index in plot_windows

[fkiraly]

reported originally by @jpss95 in #519.

Confirmed by @fkiraly on 0.21.0, current main, windows, python 3.11

---

Hi all, the same thing is happening to me with the plot_windows method. (sktime==0.18.0)

Here a code to reproduce the bug:

import pandas as pd
import numpy as np
from sktime.utils.plotting import plot_windows
from sktime.forecasting.base import ForecastingHorizon
from sktime.forecasting.model_selection import ExpandingWindowSplitter

forecast_steps= 5
initial_window_size= 40
step_length=1
fh = ForecastingHorizon(np.arange(1,forecast_steps+1))

#create dummy data
dates = pd.date_range(start='2016-01', end='2022-12', freq='M')
df = pd.DataFrame(index=dates)
df['y'] = np.random.rand(len(df))

cv = ExpandingWindowSplitter(fh=fh, initial_window=initial_window_size, step_length=step_length)

print(df)
plot_windows(cv, df)

The image clearly has wrong indexes

cc: @fkiraly

Originally posted by @jpss95 in #519 (comment)

[fkiraly]

@jpss95, I opened a new issue here, since this concerns a different plotting utility.

Thanks for reporting!