TODO StreamAdmin Klawf knife or shotgun emote (thanks R4D4R)
TODO Fun Try Minecraft
TODO StreamAdmin Expensive redeem where Crono tells lore from Vtuber chat (supported by Maweexy)
TODO StreamAdmin Add smort command
TODO StreamAdmin "Data Da Da" from Maweexy and Intelijens
TODO StreamAdmin Fix "Me with Small Screen"
TODO StreamAdmin Have custom commands in easy to reach place
TODO StreamAdmin "Crono and Costco Hot Dogs, name a more iconic duo"
TODO MeaLeon Add custom function to look at the term frequency distributions
TODO MeaLeon Research What is a good minimum document frequency for a term and what is a cutoff? 1/number of topics?
TODO MeaLeon Add ingredient information to search result box (assuming this means top n ingredients, maybe their vector value)
TODO MeaLeon Make the background tile
TODO MeaLeon take another look at Streamlit (visualizations)
TODO MeaLeon Research Look up math theory behind t-SNE
DONE MeaLeon Try Optuna for hyperparameter searching
TODO MeaLeon Add Weights and Biases for experiment logging
TODO CodeAdmin Research Should .lock and tool-versions be added to .gitignore or should they be added to the repo?
TODO MeaLeon Refactor to use Pola.rs
TODO MeaLeon Write better/more thorough docstrings
TODO Research Look up examples of applying object oriented programming practices to data science
TODO OpenSource scikit-learn Make a PR to update the documentation for how **kwargs are used for sklearn Pipeline (their example and docs are seemingly incorrect)
TODO OpenSource scikit-learn Make a PR to update the FeatureAgglomeration docs to have linkage ward also say "only euclidean is accepted"
TODO MeaLeon Add contributing guidelines to repo
TODO MeaLeon Use KMeans to predict cluster for the Missing Cuisines
:logbook:
CLOCK: [2022-12-01 Thu 23:32:02]–[2022-12-01 Thu 23:32:02] => 00:00:00
CLOCK: [2022-12-01 Thu 23:32:03]–[2022-12-01 Thu 23:32:09] => 00:00:06
:END:
Started using rawer data, and switched to HDBSCAN
Can try to compare with KMeans later
TODO MeaLeon Update READMEs
TODO MeaLeon Determine limits of free Edamam tier, may be too restrictive
TODO MeaLeon Would be cool to have something like this graph with interactivity to show closest recipes to your searched recipe
TODO CodeAdmin Link VSCode to DigitalOcean?
TODO MeaLeon switch to use TFIDF instead of OHE
DOING MeaLeon See how HDBSCAN clusters map cuisine labels
:logbook:
CLOCK: [2022-12-01 Thu 23:38:41]
CLOCK: [2022-12-01 Thu 23:38:45]
:END:
TODO MeaLeon Predict HDBSCAN clusters for recipes that are missing labels
TODO MeaLeon Research Can HDBSCAN clusters be used for label similarity exclusion? Say instead of getting a Chinese recipe and excluding all Asian food, just use HDBSCAN prediction and avoid recipes with similar/same cluster
TODO MeaLeon Further dimension reduce for a 2 or 3D visualization for plotting/sharing
TODO MeaLeon Mix in the probability to impact "density" of color (like the HDBSCAN plot examples…but in Bokeh)
