Stream notes

Video

Here’s the VOD:

Below is an embed

---

Summary

1. Intro

   1. Crono is working on call tonight
   2. Celebrate techygrrrl for 1 year anniversary as Twitch affiliate

2. Music

   1. Guest Mix: Brazilian Forró on vinyl with Worm Disco Club
   2. CBS: Jazz Records from Yugoslavia
   3. Guest Mix: Afro Cuban/Latin Jazz on Vinyl with Salt Cathedral
   4. PS: Turkish Female Singers on Vinyl with Yemeksepeti Banabi
   5. Guest Mix: Eastern European Ladies on Vinyl with Ladies on Records
   6. PS: Rare Turkish Anatolian Rock, Funk & Jazz with Yemeksepeti Banabi

3. Coding

   1. We want to improve the recipe database
   2. We need more recipes (scraping) and we need to store them somewhere (SQL or NoSQL) and we need it to run quarterly across a number of sites (possibly multiple scrapers concurrently)
   3. It looks like we can use AirFlow to schedule and automate webscraping
      1. Remember for AirFlow, things are separated into tasks and then arranged in a directed acyclic graph (DAG)
      2. Can make Docker containers and run the scraper and database adder with the containers but that would kind of imply that the entire thing would run off my desktop
         1. Would like separate device for scraper since it would be on like 24/7
            1. Maybe we can still set up Docker
         2. Maybe likewise for the database
   4. Could we use some cloud provider to offload some work? What is the cost?
   5. Variable name redeem from TomorrowsCraftsman: Potato == Yum
   6. tSNE calculation is going very slowly and not sure if something is wrong
      1. Lots of discussion about tSNE speed being not that great
         1. Like on sklearn’s own issues forum
         2. Competitor examples still slow but may be faster than sklearn
      2. Compare speeds with this tool
      3. tSNE, you’re doing it wrong, according to this article
   7. From hittaus
      1. Suggest switching to umap and/or FastTSNE
      2. Debug by doing a thousand records, time it, then another thousand, look at the increase over linear time to see the speed
      3. Cool explainer on decision trees!

4. Ucelote redeemed pun-ish me daddy for sushi

   1. "Being on call is hard, but you gotta roll with it"
   2. Sashimi more like sashi-you
   3. Maki the best of the situation
   4. What happens to sea urchins before they get turned into sushi? They go to uni-versity

5. R4D4R might start using PopOS

6. We got raided by

   1. Maweexy
   2. EroAxee
   3. Almost immediately after by LunaTopaz

---

Shoutouts

Streamers who were active in chat

1. Intelijens
2. R4D4R
3. TomorrowsCraftsman
4. Yuka_with_Data
5. Valkeryias
6. AppleGlass

---

To Do

Code

• [X] Refactor the nlp preprocessing functions into their own script to import
• [X] Change ad schedule, it’s too many ads in an hour
• [ ] Add custom function to look at the term frequency distributions
  • [ ] This is even what the scikit-learn docs do…they should just build it in to the algorithms…
• [ ] What is a good minimium document frequency for a term and what is a cutoff? 1/number of topics?
• [ ] Look at the example on sklearn for NMF and topic extraction
• [ ] add ingredient information to search result box
• [ ] Make the background tile
• [ ] Make the visualization for the cuisine map (as in use something like tSNE to see where the recipes overlap in an unsupervised fashion)
• [ ] take another at streamlit
• [ ] Look up math theory behind t-SNE
• [ ] optuna for automated hyperparameter searching

Photo

• [ ] Try to see if the preview pane in Bridge can get shifted left a little

General Stream/Admin

• [ ] How to change wordpress endpoint, specifically /admin
• [ ] Make hand cam a separate scene?
• [ ] Add new scene for 3rd monitor
• [ ] Make subgoal buying a wet fart soundboard (thanks chat)
• [ ] CornoZeewo scuffed model
• [ ] Check on the person who was asking for help with a JSON file containing weather data