Introduction to Scikit Learn Version
Scikit learn version v0.1 beta was released to the public in February 2010 by a set of developers from INRIA, the French institute for research in computer sciences, France and it was built on the initial project work done earlier by David Cournapeau as part of his Google summer project 2007, from the initial version onwards Scikit remained as a popular open-source machine learning library and it was used extensively in several real-world Artificial Intelligence (AI) projects, Scikit underwent several version changes starting from v0.1 up to v0.24 which was released as late as Apr-2021.
What is Scikit Learn?
- It is easy to develop logic and programming steps for machine learning scenarios end to end. But it is not possible to develop an ML algorithm from the scratch every time that too for a complex scenario. Scikit learn provides pre-developed codes in Python, in a library for Python developers to consume it in their program to build ML scenarios efficiently.
- These libraries provide supervised and unsupervised ML algorithms built on Technologies like NumPy, Matplotlib, Plotly, Scipy, and Pandas and cover functionalities that include several statistical models like Regression, Classification, clustering, etc. Its feature includes Dimensionality Reduction, Datasets, Parameter tuning, Ensemble methods, Supervised / Unsupervised models, Cross-validation, etc.
Various Scikit Learn Versions
Given below are the various Scikit Learn Versions:
1. Version 0.1
A beta version published by a set of developers from INRIA in Feb 2010 was an improved version of an initial google project done in 2007. This version had sponsorship from INRIA, Google, Python software foundation, Tinyclues, and 30 community contributors. It had several features like Undersampling, Oversampling, Combination of Under sampling & Oversampling, and Ensemble sampling.
2. Version 0.2
Version 0.2 released in April 2010 from Scikit Learn addressed issues in under-sampling and ensemble algorithms. Added new technique AIIKNN in under-sampling logic. The doctor was added to the documentation.
3. Version 0.3
The third release in Scikit Learns that took place in Jun 2010 fixed bugs in Check ratio functionality in the previous release. Certain functionalities in the earlier versions in datasets and dictionaries were deprecated.
4. Version 0.4
This version was released in Aug 2010 and its highlights are:
- Metrics module, GMM module was introduced.
- LARS algorithm was introduced.
- Attrselect module was obsoleted.
- Legacy codes were removed.
- New examples were introduced and many bugs were fixed.
5. Version 0.5
Introduced in Oct 2010, this release had several new classes like sparse matrices, cross-validations, efficient LARS implementation, and extraction. Documentation was improved, a new sphinx theme was introduced on the web page and bugs were fixed.
6. Version 0.6
This version was released in December 2010 with new modules like stochastic gradient descent, improved sym module, text feature extraction, and real-world data sets.
7. Version 0.7
Released in Mar 2011, this version had many performance improvement features in Gaussian model sampling, high dimensional spaces, and in many other modules. It optimized collinearity functions and refactored several algorithms for better stability.
8. Version 0.8
This version was released in May 2011 and it highlights are:
- New models such as Hierarchical clustering, Kernel PCA, Cross decomposition, non-negative matrix factorization were introduced.
- Many other modules were fine-tuned for better performance.
- Memory leaks issue and other bugs were fixed.
9. Version 0.9
Introduced in Sep 2011, it had many new modules introductions like manifold learning, Dirichlet process, performance fine-tuning, and improvement in documentation.
10. Version 0.10
This 10th version of Scikit was released in Jan 2012, supported by Python 2.6, and support to earlier versions was dropped. New modules were introduced in Ensemble methods etc. Several modules were refactored for better performance. Memory leak in the support vector method was arrested.
11. Version 0.11
Introduced in May 2012, this revision improved regression and classification functionalities. It had brought several performance improvements in API modules.
12. Version 0.12
This version was released in Sep 2012. It added Huber and Quantile functionalities in regression modules. Performance enhancements in decision trees were made. It had a sub-release 0.12.1 in Oct 2012.
13. Version 0.13
Released in Jan 2013, this version had several changes implemented in the Estimator class and API areas. To fix the bugs a sub-release 0.13.1 was released in Feb-2013.
14. Version 0.14
Introduced 7 months later in Aug 2013, this new version added bi-clustering logic and new functions in ensemble sampling. It also restructured decision trees for better performance.
15. Version 0.15
This version was released in Jul 2014 and it had several improvements in performance by fine-tuning memory usages in several functions. Shorthand constructors were added. Two sub releases 0.15.1, 0.15.2 were introduced in Aug-2014/ Sep 2014 solely to fix bugs.
16. Version 0.16
Scikit released its sixteenth version in Mar 2015. Integration with Panda frames is the new feature in this release. Performance improvement is achieved through better memory management in several modules and revisiting default settings. A subversion 0.16.1 to arrest bugs was released in Apr 2015.
17. Version 0.17
Released in Nov 2015, this version focused on new modules and enhancements. A new class was introduced in Ensemble sampling. A subversion 0.17.1 was released in Feb 2016 to fix the bugs.
18. Version 0.18
Scikit released the 18th revision in Sep 2016. It is the last release for Python 2.6. Select modules were enhanced. Two subversions 0.18.1, 0.18.2 were released in Nov 2016, Jun 2017.
19. Version 0.19
Introduced in Aug 2017 this release had significant features in classifiers, regressors, and estimators. Two sub releases 0.19.1 in Oct 2017 and 0.19.2 in Jul 2018 addressed bugs.
20. Version 0.20
Released in Sep 2018 this release focused on improvements in Library, documentation, and examples. Subversions in 0.20.1 in Nov 2018, 0.20.2 in Dec 2018, 0.20.3 in Mar 2019 and 0.20.4 in Jul 2019 were released to fix issues.
21. Version 0.21
Version 0.21 was released in May 2019 with a total change in the various models and sub releases 0.21.1, 0.21.2,0.21.3 happened in May 2019 to Jul 2019 period.
22. Version 0.22
This version was released in Dec 2019 that gave a look to its website. Two sub releases 0.21.1 and 0.21.2 were released Jan 2020, Feb 2020
23. Version 0.23
Released in May 2020 with several changes to improve the performance. Sub releases 0.23.1, 0.23.2 were released in May 2020.
24. Version 0.24
The latest version in Scikit was released in Dec 2020 that fixed many bugs. Subversions 0.24.1 were released in Jan 2021 and 0.24.2 was released in Apr 2021.
Since the future is going to be driven by AI features, Scikit Learn libraries will play a major role in the software development of machine learning applications.
This is a guide to Scikit Learn Version. Here we discuss the introduction, what is scikit learn? and various versions respectively. You may also have a look at the following articles to learn more –