TOP 10 Python Libraries You Could Have MISSED

  • by
TOP 10 Python Libraries You Could Have MISSED

Today I’m going to introduce to you 10 Python libraries that you might not have heard about.

TOP 10 Python Libraries You Could Have MISSED


The first library is Hug. Hug’s trying to make developing Python APIs as simple as possible. It’s ideal for private projects when you want to get a working API in minutes. Actually you can create a simple endpoint in just 3 lines of code. That’s a pretty good result! If you’re not familiar with backend and you’d like to run your Machine Learning API in a really short period of time – it’s a great library for you.


The second library is Bokeh. If you’re looking for a library that’s 100% independent from Matplotlib and has a full variety of possibilities – Bokeh is the right choice. Its biggest advantage is interactivity. They are moving against better health. They are improving there. All the green Latin American countries, they are moving towards smaller families. Yellow ones here. All the Arabic countries and they get larger families.

You can run scalable visualizations in a modern browser similar to Data-Driven Documents with the help of JavaScript widgets. Definitely give it a try if you haven’t yet.

CatBoost library

Number 3, is CatBoost library. It’s a high performance, scalable gradient boosting library. I’m pretty sure you’ve heard about XGBoost. CatBoost is its younger brother with an emphasis on quality. Machine Learning is a wide range of algorithms that learn from data to make better decisions and predictions. It is used in all kinds of tasks. From music and shopping recommendations to image and voice recognition to weather prediction. We have developed an open sourced, a new Machine Learning tool that provides top accuracy for the most common business cases for Machine Learning. The tool is called CatBoost.

It’s well known because of its inference time capabilities. It’s able to work on multiple GPUs as well, and you can run both regression and classification problems with it. Many modern Data Science competitions were won by using this library.


The fourth one is Eli5. It’s a tool to debug Machine Learning classifiers and help you understand their predictions. It works perfectly with most of the modern libraries like scikit-learn, Keras, XGBoost, and so on.

Moreover, it has implemented several algorithms for inspecting black-box models. It can visualize the most important features both on images and text data. Be careful of using it anytime soon, because I have read on their GitHub repository that their current build has failed.

Now, I’d like to share with you the fifth library – StatModels. It’s quite likely you haven’t heard about it but it has over 5000 stars on Github and 219 people working on it. It’s a supplement to scipy for statistical computations including descriptive statistics, estimation, and statistical models. It has multiple different models built in it. Feel free to go through the documentation. I’m pretty sure you’ll find something useful for you.


The sixth library is exceptionally interesting as it covers many different areas of Machine Learning. It’s called Pattern. It can help you in Data Mining, Natural Language Processing, Unsupervised Learning, and Network Analysis like graph centrality and visualization. I’d say this library is the whole package.

In my opinion, it’s not tested enough for commercial use, but it can absolutely be a lot of fun to play with, since you can use one library for both data mining and machine learning. It’s also well documented. Good luck in exploring it.


The seventh library has been built with Natural Language Processing in mind – it’s called Gensim. It has over 10000 stars on Github and 335 people working on it regularly. It’s built only in Python, which is its distinguishing feature. You can use it for topic modeling, document indexing, and similarity retrieval with large amounts of data.

It doesn’t require huge amounts of RAM since its algorithms are memory-independent, which means they don’t need to store lots of information in memory while performing particular jobs. We can say this library is a Python masterpiece and I’m sure you will hear about it again pretty soon.


The next library I’d like to present to you is Gluon. It’s a very carefully-prepared library by such giants as Amazon and Microsoft. As you can imagine, it’s been built to simplify the use of AWS and Azure cloud platforms. It’s been developed with developers in mind and it’s very fast and consistent.

Gluon might be a gateway between your computer and the powerful hardware used by the biggest tech giants.


Number 9: another library I’d like to propose to you is spaCy. It’s another Natural Language Processing Library and its superpower is that it was designed for multiple languages like English, German, Portuguese, French, and so on.

spaCy is not intended for academic usage. It’s a production-ready package, which means all of the algorithms are as efficient as they can be, and everything is well-tested. You will not find accidental updates here. This library has been created to be reliable. It has tokenizers and Named Entity Recognizers for all of the supported languages.

It can be easily used with TensorFlow and many other deep learning frameworks.


And last but not least, the library – Coach! It’s probably my favorite in terms of usage. With this library, you can train state-of-the-art Reinforcement Learning algorithms using multiple different games.

It has a huuuge number of supported algorithms. It was developed by Intel, so it’s quite likely you’ll hear about this library in the future. It has state-of-the-art visualization tools which can help you analyze your training results.

Overall, I do recommend trying out this package. And leave a comment telling us what your favorite Python library is – we might have missed it!


Useful links:

reference – TOP 10 Python Libraries You Could Have MISSED

Share this post ...

Leave a Reply

Your email address will not be published. Required fields are marked *