Ariel Cedola

4/4 The trained binary classification model (LightGBM) predicts the occurrence of a specific machine stop (mechanical failure) in the short-term, with very promising metrics. #MachineLearning #ArtificialIntelligence #FeatureEngineering #PredictiveMaintenance

3/4 In this work, we propose a method to decompose two streams of production and stops data from an industrial machine into a series of events, and then apply a feature engineering scheme, adapted from log-based predictive maintenance approaches, to prepare data for modelling.

I’d like to share the last paper I wrote in collaboration with my colleagues of the IoT and Robotics Research Area during my recent experience at @LinksFoundation. The paper has just received one of the Best Paper Awards at the Data Analytics 2021 Conference. 1/4

@rasbt I usually use !conda info, mainly to keep track of the python env. But this looks great, thanks for sharing @rasbt.

Wow, just realized that Watermark (IPython/Jupyter magic) turned 8 years old! Also just released a new version (github.com/rasbt/watermark), and thx to contribs, you can now print your Conda env. Hah, managing notebooks for research, teaching, & book, I wish we added this earlier 😅

My advice to new #DataScience practitioners: learn #SQL before #pandas and #mlflow before start using #MachineLearning cloud platforms.

@ChristophMolnar @trello @TiddlyWiki With GTD I learnt the importance of writing all I have in mind as a first step to get things done, great book.

My biggest improvements in self-organization: -> Zealous use of my calender -> Implementation of a version of Getting Things Done (GTD) with @Trello -> Serious note-taking (aka second brain) with @TiddlyWiki

@jrieke @streamlit Thanks for sharing these so useful advices @jrieke, and congrats for joining @streamlit!

How to get ⭐ on Github? Working on open-source projects got me >1k Github stars, tripled my Twitter followers, and landed me a fantastic job @streamlit. Here are 3 things I learned 🧵👇

Machine learning pipelines

@amuellerml @scikit_learn @sklearn_inria Great! Thanks for sharing!

The @scikit_learn team at @sklearn_inria has created a MOOC for getting started with scikit-learn! Go check it out! inria.github.io/scikit-learn-m… You can leave feedback and suggestions (or make PRs) on the repo here: github.com/inria/scikit-l…

@AzureSupport Done, here's the link: docs.microsoft.com/en-us/answers/… Thanks! ;)

@arielcedola That is a really good question! We would like to get our community's thoughts on this. Could you create a post here: msft.it/6016VbB4K with the 'azure-machine-learning' tag and share the link with us? We can escalate it internally to be seen by one of our experts! ^ED

@azuresupport #azTechHelp Hi, is there any way to access to the dataset transformed by automatic featurization steps in Azure Automated ML?

@ampanmdagaba I switched to assign some time ago, but option 1 works also well for me.

A famous #Python maxim says: "There should be one, and preferably only one, obvious way to do it". I know 6 ways to add a new formula-based column in #Pandas. Here's a comparison of times it takes (100 times on 1e6 long columns). Any advice on which one to use?

I find @streamlit extremely useful also for education purposes in #machinelearning. This is the ultra-simple #python web app I developed to teach clustering concept and k-means to my students. share.streamlit.io/arielcedola/km…

@rasbt @packt By the way @rasbt, great book 😉

Some people asked me if it is possible to get a discount on Python Machine Learning (3rd Ed) for the new semester. I don't have any codes myself, but @Packt added it to the New Year New Skills Sale on Amazon with 20% off until Monday: amazon.com/stores/page/8C…

Data scientists using Azure Synapse Analytics out there? thoughts? #Synapse #Azure #Microsoft #DataScience #MachineLearning

@chrisalbon The future of learning to read & write

I'm really enjoying reading this book about model exploration and explainability by Przemyslaw Biecek and Tomasz Burzykowski, still a work in progress but worth publishing. Explanatory Model Analysis ema.drwhy.ai #rmarkdown #bookdown @smarterpoland

@rasbt @rasbt have you not tried Mendeley Desktop for papers organization? I fully recommend it ;)

"How I Keep My Projects Organized" -- After exchanging notes on project management approaches with friends and colleagues privately & on Twitter today, I just wrote a quick blog post outlining my approach, which may or may not be useful to others :) sebastianraschka.com/blog/2021/proj…

@rasbt @rjurney Completely agree, I never felt comfortable working with notebooks e.g. on VSC. And I always preferred Notebook to Lab, but this new release of Lab is so compelling.

@rjurney I use VS Code ~80% of the time when I am writing Python code (.py files), but when working with Jupyter notebooks, I think Jupyter Lab is still the way to go. I dunno, but working with Jupyter notebooks outside Jupyter Notebook/Lab always felt limiting or buggy.

This is great. I am especially thrilled about the visual debugger.