Skip to main content

A general package to handle nested cross-validation for any estimator that implements the scikit-learn estimator interface.

Project description


This repository implements a general nested cross-validation function. Ready to use with ANY estimator that implements the Scikit-Learn estimator interface.


Be mindful of the options that are available for NestedCV. Some cross-validation options are defined in a dictionary cv_options:

Single algorithm

Here is a single example using Random Forest

# Define a parameters grid
param_grid = {
     'max_depth': [3, None],
     'n_estimators': [100,200,300,400,500,600,700,800,900,1000],
     'max_features' : [50,100,150,200] # Note: You might not have that many features

# Define parameters for function
# Default scoring: RMSE
nested_CV_search = NestedCV(model=RandomForestRegressor(), params_grid=param_grid , outer_kfolds=5, inner_kfolds=5, 
                      	    cv_options={'sqrt_of_score':True, 'randomized_search_iter':30}),y=y)
print('\nCumulated best parameter grid was:\n{0}'.format(nested_CV_search.best_params))

Multiple algorithms

Here is an example using Random Forest, XGBoost and LightGBM.

models_to_run = [RandomForestRegressor(), xgb.XGBRegressor(), lgb.LGBMRegressor()]
models_param_grid = [ 
                    { # 1st param grid, corresponding to RandomForestRegressor
                            'max_depth': [3, None],
                            'n_estimators': [100,200,300,400,500,600,700,800,900,1000],
                            'max_features' : [50,100,150,200]
                    { # 2nd param grid, corresponding to XGBRegressor
                            'learning_rate': [0.05],
                            'colsample_bytree': np.linspace(0.3, 0.5),
                            'n_estimators': [100,200,300,400,500,600,700,800,900,1000],
                            'reg_alpha' : (1,1.2),
                            'reg_lambda' : (1,1.2,1.4)
                    { # 3rd param grid, corresponding to LGBMRegressor
                            'learning_rate': [0.05],
                            'n_estimators': [100,200,300,400,500,600,700,800,900,1000],
                            'reg_alpha' : (1,1.2),
                            'reg_lambda' : (1,1.2,1.4)

for i,model in enumerate(models_to_run):
    nested_CV_search = NestedCV(model=model, params_grid=models_param_grid[i], outer_kfolds=5, inner_kfolds=5, 
                      cv_options={'sqrt_of_score':True, 'randomized_search_iter':30}),y=y)
    model_param_grid = nested_CV_search.best_params

    print('\nCumulated best parameter grid was:\n{0}'.format(model_param_grid))

cv_options documentation

metric : Callable from sklearn.metrics, default = mean_squared_error

      A scoring metric used to score each model

metric_score_indicator_lower : boolean, default = True

      Choose whether lower score is better for the metric calculation or higher score is better.

sqrt_of_score : boolean, default = False

      Whether or not if the square root should be taken of score

randomized_search : boolean, default = True

      Whether to use gridsearch or randomizedsearch from sklearn

randomized_search_iter : int, default = 10

      Number of iterations for randomized search

recursive_feature_elimination : boolean, default = False

      Whether to do feature elimination

How to use the output?

We suggest using the best parameters from the best outer score with your full data in a GridSearch Cross-Validation. They can be accessed on the NestedCV object by .best_params


  • XGBoost implements a early_stopping_rounds, which cannot be used in this implementation. Other similar parameters might not work in combination with this implementation. The function will have to be adopted to use special parameters like that.
  • The function only works with Pandas dataframes, and does currently not support NumPy arrays.
  • Limited feature selection/elimination included (only executed after inner loop has run)

What did we learn?

  • Using Scikit-Learn will lead to a faster implementation, since the Scikit-Learn community has implemented many functions that does much of the work.

Why use Nested Cross-Validation?

Controlling the bias-variance tradeoff is an essential and important in machine learning, indicated by [Cawley and Talbot, 2010]. Many articles indicate that this is possible by the use of nested cross-validation, one of them by Varma and Simon, 2006. It has many applications and has many applications. Other interesting literature for nested cross-validation are [Varoquaox et al., 2017] and [Krstajic et al., 2014].

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

nested_cv-0.913.tar.gz (3.6 kB view hashes)

Uploaded Source

Built Distribution

nested_cv-0.913-py3-none-any.whl (4.1 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page