ISSUE 439 · June 6, 2023

Projects

Later this month, Data Elixir will start releasing data science chatbots each week to serve as assistants for Data Elixir readers.

The bots use the posts, articles and papers in each newsletter as starting points and from there, they're trained on related materials so they can carry on smart discussions, help with practical applications, go deeper if you want, explain concepts that you're not sure about, answer questions about related concepts, create study guides and more. 

The stack is GPT-4 | LangChain | Pinecone, which works great for this. 

A paid subscription for the bots will be available soon and, in the meantime, I'm looking for beta testers. If you're interested in helping out, sign up here >>

Sponsored Link

Webinar | Scaling Data for Startup Success: A fireside chat

Benn and Peter will discuss the challenges that startups face when it comes to building a data-driven culture, identifying the right metrics to track, and choosing the right tools and technologies to support their data infrastructure. Save your seat.

Posts & Tutorials

17 Data Science Podcasts to Listen to in 2023

The results from last week's "Favorite Data Podcasts" poll were all over the map. 34 different podcasts were identified as being reader favorites and none had a clear majority. For identifying worthwhile data podcasts, this post from Coursera is the guide we should have started with.
Coursera

Why AI Will Save the World

There's a full-blown panic about AI right now but the real risk is losing the race to global AI technological superiority. In his latest post, Marc Andreessen explores the reasons to be excited, not scared, about AI.
a16z | Marc Andreessen

How to make fancy road trip maps with R and OpenStreetMap

This step-by-step tutorial shows how to get geocoded location and routing data from OpenStreetMap and then make fancy maps using ggplot2. Along with the road trip ideas, this is a great tutorial with lots of code snippets, screenshots, and customizations along the way.
Andrew Heiss

Dependency Management

Dependency management helps ensure that your code runs smoothly and correctly. In this post from The Missing Semester of Your DS Education, Matt Kaye shows how tools like R's renv and Python's Poetry help you declare dependencies, resolve conflicts, and isolate dependencies that are specific to each project.
Matt Kaye

Google Advanced Data Analytics Certificate

The Google Advanced Data Analytics Professional Certificate is a 7-course series that focuses on building regression and machine learning models, applying statistical methods to investigate data, creating data visualizations, and communicating insights from data analysis to stakeholders. The course is run by Coursera and is free to get started.
Data Elixir Partner

Papers

What are embeddings?

Awesome deep dive on machine learning embeddings. This 82 page paper covers what embeddings are, the problems they solve, how they work, and ways to implement them. It’s organized in a way that makes it easy to jump in wherever makes sense for you. Also, be sure to check out the "Next" tab on the intro post for worthwhile resources.
Vicki Boykis

Let's Verify Step by Step

This paper shows how training a model's "thinking" process, step by step, works substantially better than just rewarding the final outcome. The application here is mathematics and the technique also has the effect of training the model to produce chain-of-thought reasoning. 
Hunter Lightman, Vineet Kosaraju et al.

Resources

A First Course in Causal Inference

These lecture notes are based on the author's Causal Inference course at UC Berkeley. The course is designed as an introduction and covers a variety of topics from the Yule-Simpson paradox through potential outcomes, mediation and more.
Peng Ding

The {marginaleffects} Book 

The marginaleffects package for R enables users to compute and plot predictions, slopes, marginal means, and comparisons (contrasts, risk ratios, odds, etc.) for over 80 classes of statistical models. This new online book is full of tutorials, case studies, tips, and technical notes to help you get the most from the marginaleffects package. 
Vincent Arel-Bundock

Outlier

QR codes with Stable Diffusion and ControlNet

A student in Beijing figured out how to use ControlNet with Stable Diffusion to make working QR Codes! The link goes to a Chinese website that walks through a workflow with examples. Use a browser with built-in translation and it looks like a great rabbit hole.
A Zhihu Creation

How did you like this issue of Data Elixir?

👎    1     2     3     4     5   👍

Was this email forwarded to you?  Sign up here >>