Hi there!

Welcome to the third edition of the newsletter. This newsletter contains various articles ranging from geospatial data analysis to strategies for speeding up your traditional machine learning models to time series visualizations. Additionally, we’ll dissect an interesting paper from Twitter and understand some nuances of public speaking.

Geospatial data refers to the time-based data related to a specific location on the Earth’s surface. It is helpful since it can reveal vital patterns and trends in the landscape. In this article, we’ll learn about an open-source tool called Kepler.gl is and how it makes visualizing and analyzing geospatial data a seamless task. I have also made the data available as a Kaggle dataset.

This article initially started as a Linkedin post, but then I decided to make it more elaborate. Essentially, it talks about the read_clipboard()method of pandas for instantly creating a dataframe from the data copied to the clipboard.

This article is about speeding up your traditional machine learning libraries like scikit-learn by converting them to tensor-based models so that they can utilize the hardware accelerators (e.g., GPUs), thereby speeding up the model training and inference. The code is also available as a Kaggle Notebook.

Heatmap is an important EDA tool you’ll encounter on many occasions. For instance, Github’s or Kaggle’s contribution plots are also a heatmap displaying time-series data. In this article, I tried to convert vanilla time series plots into professional-looking Github-style contribution plots, which was fun.

Instead of publishing a single interview this month, I compiled all the previous interviews into a single document to have them all in a single place. I plan to add more interviews to the list as time progresses.

Some of these interviews are a couple of years old(for instance, Bojan and Rohan are now Grandmasters in all four categories), and I intend to update them to reflect the latest rankings and status. However, what doesn’t get old is their golden advice and experience in tackling some of the most challenging problems in data science.

In 2020, there was much furor on Twitter over the biased nature of their image cropping algorithm. The Twitterati complained that it was biased towards white-colored individuals and was objectifying women’s bodies. Twitter promised to look into this issue and several others to ensure responsible AI practices. This article summarizes the issues with Twitter’s Image Cropping algorithm, the findings of their research team, and the changes that followed.

Creating visualizations is easy but creating compelling visualizations is hard. In this respect, the good folks at Extreme Presentations came up with a guide called Chart Chooser to help us decide which chart to use when. You can download the freely available pdf from here.

When starting this newsletter, one of my goals was not just to share guides and tutorials but also some other interesting resources that could help you on and off your job. Here is an excellent resource on speaking confidently in public.

Communication is a valuable skill, and people who are comfortable expressing themselves generally find it easy to convey information, ideas, or even network with others. However, research shows that 85% of people are nervous when they speak in public. The nervousness heightens even more when we have to speak spontaneously.

The workshop video above is by Matt Abrahams, recorded on October 25, 2014, where he shares some techniques that can help you speak spontaneously and with greater confidence. See it in your leisure time and ponder over the things he says.

That is all for this edition. See you with another roundup next month. You can subscribe to receive the newsletter directly in your mailbox every month or share it with someone who could find them helpful.

Until next month,

Parul