Update README.md

README.md

@@ -14,5 +14,19 @@
 - output - presence of fish: proxied by presence of boats at night
 
 # Modules
-- fish presence prediction: random forest
+- fish presence prediction: gradient boosted trees
+    - yearly
+    - monthly
+    - per species/per cfa
 - data visualization
+    - core fishing areas
+    - potential fishing grounds
+
+## Training GBTs
+- training data: monthly mean environmental indicators (2017-2020), boat presence (may indicate fishing)
+- training data filtering:
+    - within 24-nm PH economic zone
+    - remove all points on shipping lines
+    - remove all points with any indicator no value
+    - remove all points with chl >= 3.5 (outlier, usually in lakes)
+    - remove all negative samples (no boat) within and near 20km of core fishing areas