View profile weekly - Issue #24: Machine Learning, faster, How to learn D3.js...


nibble dispatch

September 5 · Issue #24 · View online

Curated essays about the future of Data Science. Production Data Science and learning resources for continuous learning. Covers Data Science, Data Engineering, MLOps & DataOps. Curated by people at

This week’s top pick is Neal Lathia’s Machine Learning, faster.
Speed matters. This goes beyond thinking about minimum viable products (and the ML equivalent of “use a logistic regression before you use a neural network”); this is about the speed of the entire lifecycle for building machine learning systems.

Product Management in the era of AI
Product Management in the era of AI
5 Machine Learning Lessons for Product Managers
Some thoughtful advice on how to integrate ML into your product.
ML takes time and effort to build into your product. You need good people, good data and a good number of iterations to achieve sufficient quality – sometimes it might take over a year of work or even more. Is it something you’re happy to invest in, or would a simple, more basic heuristic be enough?
Business analytics is ridden with confirmation bias
Confirmation bias helps nobody, especially not the business or organization that suffers from it.
A lack of openness to alternative perspectives increases the chance of disagreement among decision makers with entrenched points of view, which in turn increases the chance of erroneous decisions or no decisions at all (decision paralysis).
Advanced analytics is nice, but how about we start with simple analytics?
Advanced analytics! Everyone wants some, but very few people need some. How about starting with a healthy dose of simple analytics and we’ll go from there?
Before going through the motion of hiring people with fancy degrees and expensive resumes, you need to ask yourself and your company simple analytics questions like, can we even access data to do advanced analytics on?
How organizations are sharpening their skills to better understand and use AI
To successfully implement AI technologies, companies need to take a holistic approach toward retraining their workforces.
AI important to a company’s success (EY study)
But lack of talent remains a major hurdle. Along with a lack of a clear business case for the technology.
Google just released Neural Structured Learning
An open-source framework for training deep neural networks with structured signals.
It implements Neural Graph Learning, which allows training neural networks using graphs. You can check the release post on medium or check the repo on github.
Koalas by Databricks
The Koalas project makes data scientists more productive when interacting with big data, by implementing the pandas DataFrame API on top of Apache Spark.
Learning resources
How to learn D3.js
An interactive guide for learning D3.js
D3.js is notoriously difficult to get started with. You need an understanding of many different concepts, and the library itself is composed of multiple modules. This interactive blog post groups concepts by function, and helps you navigate the library.
The blog post itself is using D3.js and the source code is available here.
Activation Functions Explained
An extensive article that goes over 6 different activation functions, each with pros and cons, equations, differentiated equations, and plots.
Deploying a Scikit-Learn model with ONNX and FastAPI
The goal of this article is to give you an introduction to ONNX Runtime and FastAPI.
ONNX is an open-source model standard that allows exchanging models between different frameworks. It enables us to train models with any kind of framework as long as the frameworks are supporting ONNX. In my opinion, this is a game-changer because data scientist can use their favorite tool to train a model while the machine learning engineers only have to set up one production environment that can run the model.
Generalized Huber Regression
This post presents a generalized version of the Huber loss function which can be incorporated with Generalized Linear Models (GLM) and is well-suited for heteroscedastic regression problems.
Did you enjoy this issue?
In order to unsubscribe, click here.
If you were forwarded this newsletter and you like it, you can subscribe here.
Powered by Revue