Dendrology

Where data meets insight → charting a clearer future

Environmental Data Scientist

From "what is an algorithm" to a working, evaluated weather dataset: the machine-learning wing's entries sequenced into one route, with checkpoints that turn reading into practice.

Who it's for: Readers starting from little or no machine-learning background who want to work with environmental data — no prerequisites beyond curiosity and a computer that runs Python.

8 of 10 steps available today — 5 published entries + 3 checkpoints; 2 entries still planned

This path is the live, free portion of the planned Environmental Data Scientist mastery track — the reading route exists today; the full program sketched there remains a design draft.

What you'll be able to do

Enroll & track progress

Everything below is readable without an account. Signing in adds memory, not walls: your enrollment and checkmarks, nothing more.

Sign in to enroll

The steps

  1. Survey the Machine Learning wing published

    The lay of the land — what algorithms are, how learning systems differ, and the vocabulary the rest of the path leans on.

  2. Genetic Algorithms published

    The archive's flagship optimization entry — evolution as a search strategy.

  3. Heuristic Models published

    The counterweight to genetic search — workable answers fast, when speed beats perfection.

  4. Checkpoint: pick the right tool checkpoint

    Take one concrete problem — say, scheduling irrigation across five fields with a limited water budget — and write half a page on whether you'd reach for a genetic algorithm or a heuristic first, and what would make you switch. There's no single right answer; the point is arguing the trade-off in your own words.

  5. Importing from APIs published

    Getting real weather data flowing — the restaurant-menu model of ordering exactly the fields you need.

  6. Writing CSV Files published

    Storing what you fetched so every tool can read it back — and the header, type, and timezone traps to avoid.

  7. Checkpoint: build your first dataset checkpoint

    Using the two entries above, pull 30 days of hourly temperature and precipitation for your own location from Open-Meteo, write it to CSV, then reload it and confirm the types and timestamps survived the round trip. Keep the file — later steps reuse it.

  8. Time-Series Basics for Weather Data planned

    Why weather rows can't be shuffled — the planned entry that closes the gap between having data and modeling it.

    This entry hasn't been written yet — it's on the archive's writing backlog, and this step will open the moment it publishes.

  9. Model Evaluation and Validation planned

    The planned capstone entry — how to know whether a model actually works before trusting it.

    This entry hasn't been written yet — it's on the archive's writing backlog, and this step will open the moment it publishes.

  10. Checkpoint: design an evaluation checkpoint

    Sketch how you would judge a 7-day temperature forecast for your area: what naive baseline it must beat, which error metric you'd track, and how you'd split past data so the test never sees the future. One page, no code required.

Step availability is derived from the archive at build time: 5 published entries, 3 checkpoints, 2 planned entries.