Video Link:
Here’s the VOD: https://www.twitch.tv/videos/1486167650
Summary:
- Brief talk about keyboards
- My keyboard is a Reduced Ergodox made by Falbatech: https://github.com/mattdibi/redox-keyboard
- Firmware for the keyboard here: https://github.com/mattdibi/qmk_firmware/tree/master/keyboards/redox_w
- Looking into replacement keyboards like the Iris: https://keeb.io/collections/iris-split-ergonomic-keyboard/products/iris-keyboard-pre-built
- If you’re doing time series forecasting, take a look at Prophet: https://facebook.github.io/prophet/ which has drastically improved their quality of documentation (in Python)
- Talked about MeaLeon again, and Heroku was actually running!
- Git repo: https://github.com/AaronWChen/MeaLeon
- Webapp: https://mealeon.herokuapp.com/
- Career prep resources page demo:
- Go to Helpful Data Science Career Resources
- For stats, free textbook here: https://www.statlearning.com/online-course
- https://www.reddit.com/r/datascience/comments/syjt0c/hopefully_almost_everything_you_need_to_know/
- Need to add levels.fyi: https://www.levels.fyi/comp.html?track=Data%20Scientist®ion=807
- Food: like hot sauce? https://heatonist.com/
- Statsmodels library for Python https://www.statsmodels.org/devel/user-guide.html
- However, the documentation is hard to follow for someone reading just the code calls since not every function or method is fully specified in those specific method docs.
- Usually have to use a mix of tutorials (some not even from statsmodels) with documentation to figure out how to do things
- Commented on how R as a language since it was intended to be a statistics package/software is just much better at doing statistics. Statsmodels introduced a formula api that basically just allows you to write R formulas and they just work. Example here: https://www.statsmodels.org/devel/example_formulas.html notably, the formula documentation is not quite as good
- Possibility of making a Quarto template and/or switching to Quarto https://quarto.org/
- Natural Language Processing (NLP) libraries:
- If you want to try NLP stuff, the most reliable/oldest is Natural Language Toolkit: https://www.nltk.org/
- In production, people seem to have switched to spaCy (https://spacy.io/) which I use
- Also Stanza https://stanfordnlp.github.io/stanza/
- And Snorkel https://snorkel.ai/
- If using Spark, check SparkNLP https://nlp.johnsnowlabs.com/
- Take a look at stopwords, which are the human expert determined words that basically provide no value and are thus excluded from tokenized list: https://gist.github.com/sebleier/554280
- Looking for advice on how to structure code?
- I recommend using Google Style Guides: https://google.github.io/styleguide/pyguide.html
- Chat submitted/related things:
- Handling geodata: geopy https://geopy.readthedocs.io/en/stable/
- Debian packaging for fedorans: Debian Packaging for Fedorans | garbage-collector-gnk
- Pixel art animation process: aarthificial 2 – YouTube
To Do:
Figure out how to make a template for blog posts on this site- Cannot be done on free tier
- Consider adding Quarto support/template
- Quarto documentation can be found here: Quarto
- Figure out how the experiment logging in DVC works
- Does it require architecture/infrastructure set up like MLFlow prefers?
- How difficult would it be to migrate from local to cloud?
- Is it better to start in cloud? How much would that cost?
- Test the Contact Me form on the Contact page/set up email to work on BuenosDS.dev
- On stream, I commented that the site doesn’t have email functionality yet, but afterwards, I realized that WordPress I believe defaults to storing anything submitted via Contact Us form into a saved form that can be accessed by site admins. Will have to check
- Figure out how to properly embed Twitch VODs in WordPress posts
- Wanted to display it on each note page, but it seems to break. Embeds mention looking at developer docs.
- Other sites mention using different blocks or features that don’t seem to exist anymore
Add stuff from chat, and also check what chat sends in for stream notes