"""Session 16 (extra lab) - two classic agricultural classification datasets.

A) UCI Rice (Cammeo vs Osmancik) - shipped in datasets/rice_cammeo_osmancik.csv (CC BY 4.0)
   Cinar & Koklu (2019). https://archive.ics.uci.edu/dataset/545/rice+cammeo+and+osmancik
B) Crop Recommendation (N, P, K, temperature, humidity, ph, rainfall -> crop)
   Download Crop_recommendation.csv from Kaggle (free account):
   https://www.kaggle.com/datasets/atharvaingle/crop-recommendation-dataset
   and put it in datasets/ - the script skips part B if the file is missing.
"""
import pandas as pd
from sklearn.ensemble import RandomForestClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score, confusion_matrix
from sklearn.model_selection import cross_val_score, train_test_split
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import StandardScaler

from common import DATA

# ---------------- A) Rice
rice = pd.read_csv(DATA / "rice_cammeo_osmancik.csv")
X, y = rice.drop(columns="Class"), rice["Class"]
X_tr, X_te, y_tr, y_te = train_test_split(X, y, test_size=0.25, stratify=y, random_state=0)
for name, m in {"LogisticRegression": make_pipeline(StandardScaler(), LogisticRegression(max_iter=500)),
                "RandomForest": RandomForestClassifier(n_estimators=200, random_state=0)}.items():
    m.fit(X_tr, y_tr)
    print(f"[Rice] {name:18s} test accuracy = {accuracy_score(y_te, m.predict(X_te)):.3f}")
print(confusion_matrix(y_te, m.predict(X_te)))

# ---------------- B) Crop recommendation
path = DATA / "Crop_recommendation.csv"
if not path.exists():
    print("\n[Crop] Crop_recommendation.csv not found - download it from Kaggle (see docstring).")
else:
    crop = pd.read_csv(path)
    Xc, yc = crop.drop(columns="label"), crop["label"]
    rf = RandomForestClassifier(n_estimators=300, random_state=0)
    scores = cross_val_score(rf, Xc, yc, cv=5)
    print(f"\n[Crop] 5-fold accuracy = {scores.mean():.3f} ± {scores.std():.3f}")
    rf.fit(Xc, yc)
    print(pd.Series(rf.feature_importances_, index=Xc.columns).sort_values(ascending=False).round(3))
    # Example: KFS farm soil test (edit with your lab results)
    sample = pd.DataFrame([{"N": 90, "P": 42, "K": 43, "temperature": 26.0, "humidity": 65,
                            "ph": 7.8, "rainfall": 30.0}])
    print("Recommended crop for the sample soil:", rf.predict(sample)[0])