The Best Machine Learning Frameworks & Extensions for Scikit-learn

Plenty of packages implement the Scikit-learn estimator API.

If you’re already familiar with Scikit-learn, you’ll find the integration of these libraries pretty straightforward.

With these packages, we can extend the functionality of Scikit-learn estimators, and I’ll show you how to use some of them in this article.

Data formats

In this section, we’ll explore libraries that can be used to process and transform data.


You can use this package to map ‘DataFrame’ columns to Scikit-learn transformations. Then you can combine these columns into features.

To start using the package, install ‘sklearn-pandas’ via pip. The ‘DataFrameMapper’ can be used to map pandas data frame columns into Scikit-learn transformations. Let’s see how it’s done.

First, create a dummy DataFrame:

The `DataFrameMapper’ accepts a list of tuples — the first item’s name is the column name in Pandas DataFrame.

The second passed item is the kind of transformation that will be applied to the column.

To get in-depth knowledge on ML lets try Machine Learning Books

For example, ‘LabelBinarizer’ can be applied to the ‘Uni’ column, whereas the ‘Age’ column is scaled using a ‘StandardScaler’.

After defining the mapper, next we use it to fit and transform the data.

The `transformed_names_` attribute of the mapper can be used to show resulting names after the transformation.

Passing `df_out=True` to the mapper will return your results as a Pandas DataFrame.


This package combines n-dimensional labeled arrays from xarray with Scikit-learn tools.

You can apply Scikit-learn estimators to ‘xarrays’ without losing their labels. You can also:

  • ensure compatibility between Sklearn estimators with xarray DataArrays and Datasets,
  • enable estimators to change the number of samples,
  • have pre-processing transformers.

Sklearn-xarray is basically a bridge between xarray and Scikit-learn. In order to use its functionalities, install ‘sklearn-xarray’ via pip or ‘conda’.

The package has wrappers, which let you use sklearn estimators on xarray DataArrays and Datasets. To illustrate this, let’s first create a ‘DataArray’.

Select one transformation from Sklearn to apply to this ‘DataArray’. In this case, let’s apply the ‘StandardScaler’.

Wrapped estimators can be used in Sklearn pipelines seamlessly.

When fitting this pipeline, you will just pass in the DataArray.

Similarly, DataArrays can be used in a cross-validated grid search.

For this, you need to create a ‘CrossValidatorWrapper’ instance from ‘sklearn-xarray’.

After that, you will fit the ‘gridsearch’ to X and y in the ‘DataArray’ data type.


Are there tools and libraries that integrate Sklearn for better Auto-ML? Yes there are, and here are some examples.


With this, you can perform automated machine learning with Scikit-learn. For the setup you need to install some dependencies manually.

Next, install ‘auto-sklearn’ via pip.

When using this tool, you don’t need to worry about algorithm selection and hyper-parameter tuning. Auto-sklearn does all that for you.

It does this thanks to the latest advances in Bayesian optimization, meta-learning, and ensemble construction.

To use it, you need to select a classifier or regressor, and fit it to the training set.

Auto_ViML — Automatic Variant Interpretable Machine Learning” (pronounced “Auto_Vimal”)

Given a certain dataset, Auto_ViML tries out different models with varying features. It eventually settles on the best performing model.

The package also selects the least number of features possible in building the model. This gives you a less complex and interpretable model. This package also:

  • helps you clean data by suggesting changes to missing values, formatting, and adding variables;
  • classifies variables automatically, whether it’s text, data, or numerical;
  • generates model performance graphs automatically when verbose is set to 1 or 2;
  • lets you use of ‘featuretools’ for feature engineering;
  • handles imbalance data when ‘Imbalanced_Flag’ is set to ‘True’

To see it in action, install ‘autoviml’ via pip.

TPOT — Tree-based Pipeline Optimization Tool

This is a Python-based auto-ml tool. It uses genetic programming to optimize machine learning pipelines.

It explores multiple pipelines in order to settle on the best one for your dataset.

Install ‘tpot’ via pip to start tinkering with it. After running ‘tpot’, you can save the resulting pipeline in a file. The file will be exported once the exploration process is completed or when you terminate the process.

The snippet below shows how you can create a classification pipeline on the digits dataset.

Feature Tools

This is a tool for automated feature engineering. It works by transforming temporal and relational datasets into feature matrices.

Install ‘featuretools[complete]’ via pip to start using it.

Deep Feature Synthesis (DFS) can be used for automated feature engineering.

First, you define a dictionary containing all entities in a dataset. In ‘featuretools’, an entity is a single table. After that, the relationship between the different entities is defined.

The next step is to pass the entities, list of relationships, and the target entity to DFS. This will get you the feature matrix and the corresponding list of feature definitions.


You can use Neuraxle for hyperparameter tuning and AutoML. Install ‘neuraxle’ via pip to start using it.

Apart from Scikit-learn, Neuraxle is also compatible with Keras, TensorFlow, and PyTorch. It also has:

  • parallel computation and serialization,
  • time series processing through the provision of abstractions key to such projects.

To do auto-ml with Neuraxle, you need:

  • a defined pipeline
  • a validation splitter
  • definition of a scoring metric via the ‘ScoringCallback’
  • a selected ‘hyperparams’ repository
  • a selected ‘hyperparams’ optimizer
  • an ‘AutoML’ loop

Check out a complete example here.

Experimentation frameworks

Now it’s time for a couple of SciKit tools that you can use for machine learning experimentation.

SciKit-Learn Laboratory

SciKit-Learn Laboratory is a command-line tool you can use to run machine learning experiments. To start using it, install `skll` via pip.

After that, you need to obtain a dataset in the `SKLL` format.

Next, create a configuration file for the experiment, and run the experiment in the terminal.

When the experiment is complete, multiple files will be stored in the results folder. You can use these files to examine the experiment.


The Scikit-learn integration of Neptune lets you log your experiments using Neptune. For instance, you can log the summary of your Scikit-learn regressor.

Hey This Is priya Reddy Iam a tech writer