expected-greenness-ca / predict_ca.py
Konuktor92's picture
expected greenness v2-ca
692b68d verified
Raw History Blame Contribute Delete
3.58 kB
"""TrendeX expected greenness, Central Asia (v2-ca): CPU inference. Imports only numpy, pandas, xgboost.
Files next to this module: model_ca.ubj, feature_schema.json, climatology.parquet (cell_id, month, clim_ndvi),
weather_climatology.parquet (cell_id, month, <feature>_mean), conformal.json.
from predict_ca import load
m = load("path/to/repo")
m.predict(df) # df has feature_schema.json columns
m.what_if(cells, cell_ids, month=7, precip_pct=-30, temp_c=1) # cells = ca-cube cells.parquet rows
Expected greenness from weather and terrain alone; the gap to observed NDVI is change NOT explained by
precipitation, air temperature, snow and soil moisture. Not a cause, not forage quality.
"""
from __future__ import annotations
import json
from pathlib import Path
import numpy as np
import pandas as pd
import xgboost as xgb
class ExpectedGreennessCA:
def __init__(self, root):
root = Path(root)
s = json.loads((root / "feature_schema.json").read_text())
self.features, self.static, self.weather = s["features"], s["static"], s["weather"]
self.booster = xgb.Booster()
self.booster.load_model(str(root / "model_ca.ubj"))
self.booster.set_param({"device": "cpu"})
self.clim = pd.read_parquet(root / "climatology.parquet").set_index(["cell_id", "month"])
self.wclim = pd.read_parquet(root / "weather_climatology.parquet").set_index(["cell_id", "month"])
self.conf = {int(k): v for k, v in json.loads((root / "conformal.json").read_text()).items()}
def predict(self, df: pd.DataFrame) -> pd.DataFrame:
missing = [c for c in self.features if c not in df.columns]
if missing:
raise ValueError(f"missing features: {missing}")
X = df[self.features].to_numpy(dtype="float32")
pred = self.booster.predict(xgb.DMatrix(X, feature_names=self.features))
months = df["month"].astype(int).to_numpy()
lo = np.array([self.conf[m][0] for m in months])
hi = np.array([self.conf[m][1] for m in months])
exp = df["clim_ndvi"].to_numpy() + pred
return pd.DataFrame({"pred_anom": pred, "expected": exp, "p10": exp + lo, "p90": exp + hi}, index=df.index)
def what_if(self, cells: pd.DataFrame, cell_ids, month: int, precip_pct: float = 0.0, temp_c: float = 0.0,
soil_pct: float = 0.0) -> pd.DataFrame:
ids = np.asarray(cell_ids)
key = pd.MultiIndex.from_arrays([ids, np.full(len(ids), int(month))])
w = self.wclim.reindex(key)
st = cells.set_index("cell_id").reindex(ids)
def frame(pp, tc, sp):
f = pd.DataFrame({k: st[k].to_numpy() for k in self.static})
f["month"] = float(month)
f["clim_ndvi"] = self.clim["clim_ndvi"].reindex(key).to_numpy()
for k in ("precip_m", "precip_2m", "precip_winter"):
f[k] = w[f"{k}_mean"].to_numpy() * pp / 100.0
f["tair_m"] = f["tair_prev"] = tc
f["soil_m"] = w["soil_m_mean"].to_numpy() * sp / 100.0
f["swe_apr"] = 0.0
return f
res = self.predict(frame(precip_pct, temp_c, soil_pct))
base = self.predict(frame(0.0, 0.0, 0.0))
res.insert(0, "cell_id", ids)
res["expected_typical_weather"] = base["expected"].to_numpy()
res["change_vs_typical"] = res["expected"].to_numpy() - base["expected"].to_numpy()
return res
def load(root=Path(__file__).parent) -> ExpectedGreennessCA:
return ExpectedGreennessCA(root)