Top Twitter Topics by Data Scientists #33

Trending this week: Use computer vision to spot politicians distracted by their phones; Learn how to select the best deep learning object detectors; 5 key steps to begin your career as a successful freelance data scientist.

Every week we analyze the most discussed topics on Twitter by Data Science & AI influencers.

The following topics, URLs, resources, and tweets have been automatically extracted using a topic modeling technique based on Sentence BERT, which we have enhanced to fit our use case.

Want to know more about the methodology used? Check out this article for more details, and find the codes in this Github repository!


This week, Data Science and AI influencers on Twitter have talked about:

  • Spotlight On Computer Vision
  • Making The Most Out Of Python
  • Guide To Succeed In Data Science

Spotlight On Computer Vision

This week, data science and AI influencers have shared a series of articles on object detection and its uses.

Pascal Bornet has shared an amazing post presenting an AI Tool Which Tracks the Time Politicians Spend on Their Phones. This post shows a funny application of computer vision to build a tool that calculates how much time politicians are distracted by their phones during meetings. Called The Flemish Scrollers, this tool is written in Python and uses object detection and face recognition technologies to, respectively, detect phones and identify faces in real-time. To make it even funnier, the tool sends automatically a tweet to remind those people identified using their phone for a while to stay focused.

KDnuggets have shared the following:

Deep Learning-based Real-time Video Processing 101 article. This article explores how to build a pipeline and process real-time video with deep learning to apply this approach to specific use cases. Serial and parallel video processing paradigms are presented along with their pros and cons. Then, a pipeline combining both approaches is introduced. This pipeline approach is more flexible, and its components can be manipulated with ease depending on the requirements of your use case. Finally, it allows applying computer vision techniques to deal with various tasks from video data.

An article presenting Metrics to Use to Evaluate Deep Learning Object Detectors. This post helps to understand which metric should be used to evaluate trained object detectors and which one is more important. It provides metrics to quantify models performance and decide which one is better for our use case. It introduces IoU, Precision, Recall, F1 Score, AP, and mAP. Then, it compares mAP to the other metrics. Finally, it gives recommendations on which metric you should use with regard to the requirements of your use case, and how to choose the best performing object detection model.

Making The Most Out Of Python

Some interesting posts providing utility Python packages to accomplish various tasks and to use Python to the fullest have been shared.

Mike Tamir has shared an article on converting static pandas plot (matplotlib) to interactive. The tutorial article talks about how to continue to use the existing Pandas interface for plotting interactive graphs. They have introduced a library called hvplot which provides a wrapper around pandas so that it can make use of an interactive plotting library called holoviews for plotting.

Sebastian Raschka has shared a link to access fastcore and its parallel submodule. It seems like another excellent way for making threading and multiprocessing in Python more accessible.

Fastcore uses Python’s flexibility to add to Python features inspired by other languages like multiple dispatches from Julia, mixins from Ruby, and currying, binding, and more from Haskell. It also adds some “missing features” and cleans up some rough edges in the Python standard library, such as simplifying parallel processing and bringing ideas from NumPy over to Python’s list type.

Finally, KDnuggets have shared two updates. First, they shared an article on easy SQL in native python. The article talks about SQLModel, a Python library for interacting with SQL databases in pure, native Python. Its design motivations include intuitiveness, ease of use, compatibility, and robustness. SQLModel employs Python type annotation, enforced and managed by Pydantic, as well as SQLAlchemy, “Python SQL toolkit and Object Relational Mapper,” for its SQL interaction.

They also shared an article on text mining in python along with steps and examples. The article talks about NLP and POS tagging along with code snippets in Python.

Guide To Succeed In Data Science

Some articles providing guidance on how to succeed in your data science career have been posted.

Marcus Borba has shared a post introducing the Top 5 Statistical Data Analysis Techniques a Data Scientist Should Know. This post presents the techniques that will provide a layman the statistical knowledge he should absolutely have in order to perform data analysis. It divides them into 5 groups: linear regression, classification, resampling methods, tree-based methods, and unsupervised learning. Finally, it gives the must-know methods you should master for each category.

On their side, KDnuggets:

Shared a post providing 10 resources for data science self-study. This post gives a list of resources that will help you in mastering the fundamentals of data science. This list is divided into resources for building fundamental knowledge, resources for data science practice, resources for networking, and continuous studies.

Additionally, they provided a list of 14 Data Science projects to improve your skills. These great project ideas varying from easy to advanced difficulty levels will help you to develop new skills and strengthen your portfolio. They are classified into three types: visualization projects, exploratory data analysis (EDA), and prediction modeling projects.

Finally, provided guidance on How to Succeed in Becoming a Freelance Data Scientist. This post is very useful in this particular time where finding a job that fits your needs can sometimes be a challenge. In these instances, freelance work could represent a suitable option. This post gives the five steps that will help you to begin your career as a successful freelance data scientist, namely: build a presence, develop new skills, work across industries, use online resources, and iron out the details. Each of these steps is explained in the post.