Add xgboost sampler

[AldoGl]

Proposed changes

• I added the XGBoost sampler used in "Agent-based model calibration using machine learning surrogates"
• I added a test for the new sampler
• I added the XGBoost sampler to the Jupyter notebook with the samplers overview

Types of changes

What types of changes does your code introduce?
Put an x in the boxes that apply

• Bugfix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)

Checklist

Put an x in the boxes that apply.

• I have read the CONTRIBUTING doc
• I am making a pull request against the develop branch (left side). Also you should start your branch off our develop.
• Lint and unit tests pass locally with my changes
• I have added tests that prove my fix is effective or that my feature works

[codecov-commenter]

Codecov Report

Merging #30 (f51d2bf) into main (f666b54) will increase coverage by 0.08%.
The diff coverage is 98.00%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #30      +/-   ##
==========================================
+ Coverage   97.50%   97.59%   +0.08%     
==========================================
  Files          24       28       +4     
  Lines        1325     1413      +88     
==========================================
+ Hits         1292     1379      +87     
- Misses         33       34       +1     

Impacted Files	Coverage Δ
black_it/samplers/xgboost.py	98.00% <98.00%> (ø)
black_it/__init__.py	100.00% <0.00%> (ø)
black_it/utils/base.py	100.00% <0.00%> (ø)
black_it/utils/__init__.py	100.00% <0.00%> (ø)

[marcofavoritobi]

(very picky) Could you fully execute the notebook from scratch so to make the execution count reasonable?

[marcofavoritobi]

LGTM, this is a useful feature.

I wonder whether we could develop a generalized sampler based on machine learning surrogates (i.e. regressors). Something like:

class Regressor(ABC):

    @abstractmethod
    def fit(self, X, y):
        raise NotImplementedError

    @abstractmethod
    def predict(x):
        raise NotImplementedError


class SurrogateSampler(BaseSampler):


    def __init__(  # pylint: disable=too-many-arguments
        self,
        batch_size: int,
        regressor: Regressor,
        random_state: Optional[int] = None,
        max_deduplication_passes: int = 5,
        candidate_pool_size: Optional[int] = None):
        ...
        self.regressor = regressor

[AldoGl]

Yes I completely agree, it's something that I also thought I would do, but I wanted to do this in a different PR

[AldoGl]

Thank you for the review @marcofavoritobi