AlgoCatch

Model accuracy

Production species models ranked by validated AUC — the same scores that power live habitat predictions on the map.

Sign in to use the app

Last updated

14 July 2026

Live models

6 ready to use

In training

11 calibrating

Model improving in real time as we collect more data.

How to read the score

AUC at a glance

AUC score measures how accurately the model predicts where a species will be found. 0.5 = random guess. 1.0 = perfect. Scores above 0.75 are considered reliable.

0.5

Random guess

>0.75

Reliable

1.0

Perfect

Ready to use

Live models

Ranked by production AUC — highest first.

6 species

  1. Orada

    Sparus aurata

    0.826

    Prod AUC

    0.5 chance0.75 reliable1.0
    Reliable↑ +0.090 vs Mediterranean baseline

  2. Gof

    Seriola dumerili

    0.814

    Prod AUC

    ReliableStable

  3. Rdeča bodika

    Scorpaena scrofa

    0.788

    Prod AUC

    ReliableStable

  4. Mol

    Merlangius merlangus

    0.777

    Prod AUC

    ReliableStable

  5. Squid

    Loligo vulgaris

    0.737

    Prod AUC

    PublishedStable

  6. Ribon

    Pagellus erythrinus

    0.706

    Prod AUC

    PublishedStable

Coming soon

Currently training

These species are calibrating against new data. Scores stay unpublished until they clear the production threshold — we do not show partial numbers.

  • Albacore

    Thunnus alalunga

    Calibrating

  • Bluefish (Strelka)

    Pomatomus saltatrix

    Calibrating

  • Bonito

    Sarda sarda

    Calibrating

  • Brancin

    Dicentrarchus labrax

    Calibrating

  • Cuttlefish

    Sepia officinalis

    Calibrating

  • Divji ribon

    Pagellus acarne

    Calibrating

  • Kovač

    Zeus faber

    Calibrating

  • mahi

    Coryphaena hippurus

    Calibrating

  • Oslič

    Merluccius merluccius

    Calibrating

  • Sredozemski šur

    Trachurus mediterraneus

    Calibrating

  • Zobatec

    Dentex dentex

    Calibrating