All Hands on Data #44
Version 44 of AHoD welcomes a new sailor to the team, Reed! Check out his article choice this week on hyperparameter tuning below.
Querying a Billion Rows of AWS Cost Data 100X Faster with DuckDB
Always choose the right tool for the job is essentially the point of the piece. While evaluating their cloud costs report, the Vantage team opted to switch to DuckDB over PostgreSQL and saw some pretty stunning performance improvements. - John Forstmeier
Pros and Cons of Using OpenAI in Mobile App Development
The article does a nice job of pointing out not just the pros, but the cons of using OpenAI. OpenAI is phenomenal and obviously going to change a lot of the ways Mobile Apps are developed in the future. But it is really important to keep in mind some of then drawbacks of using tools like OpenAI in app development. - Jon Davidson
NoSQL Databases and Their Use Cases
Most people in the data space will have direct experience with relational SQL databases, but NoSQL databases are not as common place. One is not better than the other, but each offer certain advantages under particular use cases. At the end of the day, it is important to use the right tool for the job, and this article provides some insight into when it might be advantageous to go the NoSQL route. - Wes Poulsen
Distributed Hyperparameter Tuning in Vertex AI Pipeline
The article shows how to implement distributed hyperparameter tuning in Vertex AI, a machine learning platform provided by Google Cloud. It’s a quick look with code examples on how to optimize and create a pipeline for distributed hyperparameter tuning using Vertex AI’s built-in tools, such as the AI Platform Training and Hyperparameter Tuning services. Although there are limitations discussed, it’s a cool overview on how to spin up some ML examples in GCP. - Reed Cowan
Reading Minds with AI: Researchers Translate Brain Waves to Images
In this incredibly fascinating article, researchers are scanning peoples brains while they think about particular images. They are then using AI to reproduce the images based on the brain waves! Of course, there are ethical concerns, but the impact this technology could have on psychology or neuroscience is impressive. - Katt Baum
Enhancing the Interoperability of Data Tools
In this article, I discuss the importance of interoperability between data tools and platforms. I argue that the lack of interoperability in the data space can lead to vendor lock-in and create significant barriers to entry for new players. The goal is to create a more open and connected ecosystem that allows for seamless data integration and facilitates innovation in the data space. Increased interoperability will not only benefit end-users but also encourage healthy competition and innovation among vendors. - Blake Burch
How to Turn Boring Visualization into Fascinating Data Storytelling.
When I taught math, I always described my teaching method as trying to weave a story through the full curriculum that makes sense as I continue to build and add things to it. I think the data space could use the same treatment. In this article, Alaa describes how to turn a dashboard into a story which I think could be great for everyone to do in their organizations. - Steven Johnson
Comparing List Comprehensions vs. Built-In Functions in Python: Which Is Better?
There paradox of choice is a great description of what goes on in the day to day when deciding ow to move forward with programming. As someone with less Python experience, this article is great at laying out major differences and considerations between list comprehension and it’s functional counterpart. - Eric Elsken