Fixed deprecation warning

[jorenham]

This fixes several cases of:

FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`

[pep8speaks]

Hello @jorenham! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! :beers:

Comment last updated at 2023-09-27 16:15:58 UTC

[codecov[bot]]

Codecov Report

All modified lines are covered by tests :white_check_mark:

Files	Coverage Δ
arch/tests/univariate/test_forecast.py	100.00% <100.00%> (ø)
arch/tests/univariate/test_mean.py	100.00% <100.00%> (ø)
arch/tests/univariate/test_rescale.py	100.00% <100.00%> (ø)
arch/unitroot/_engle_granger.py	100.00% <100.00%> (ø)
arch/unitroot/unitroot.py	99.86% <100.00%> (ø)
arch/univariate/base.py	98.93% <100.00%> (-0.02%)	:arrow_down:

:loudspeaker: Thoughts on this report? Let us know!.

[bashtage]

I think it would be better to explicitly use .iloc rather can converting the entire array.

There are quite a few in the test log. Feel like tackling the others?

[jorenham]

@bashtage I wanted to to that before, but after looking at the statsmodels LikelihoodModelResults class, I saw that params, and therefore tvalues, could also be a numpy array. This way it'll work in both cases.

I'll look into the remaining warnings as well.

[jorenham]

@bashtage I'm not sure why this test is failing. https://dev.azure.com/kevinksheppard0207/kevinksheppard/_build/results?buildId=944&view=logs&j=7cb0d55f-58f1-52d4-9308-3ef157e47daa&t=56ed1dc5-eb59-5286-8983-2e32283fa027&l=93

[jorenham]

I think it would be better to explicitly use .iloc rather can converting the entire array.

There are quite a few in the test log. Feel like tackling the others?

For series of length 2, it appears that np.asarray(params)[0] is a bit faster than params.iloc[0] (1.4µs vs 1.6µs), using pandas 2.1 on python 3.11. But I found a better solution, param0, _* = params, which runs on average in 430 ns, which also works for pandas.Series and np.ndarray objects.

So I'll go ahead and update the changes to this unpacking method.

[bashtage]

Looks like 2 final one:

rch/tests/univariate/test_mean.py::test_param_cov
  /opt/hostedtoolcache/Python/3.11.5/x64/lib/python3.11/site-packages/statsmodels/tools/numdiff.py:150: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
    ei[k] = epsilon[k]

arch/tests/univariate/test_mean.py::test_param_cov
  /opt/hostedtoolcache/Python/3.11.5/x64/lib/python3.11/site-packages/statsmodels/tools/numdiff.py:151: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
    grad[k, :] = (f(*((x+ei,) + args), **kwargs) - f0)/epsilon[k]

[bashtage]

Thanks. Going ot merge this. I'll catch the other two.