Upload 3 files

Dockerfile

@@ -0,0 +1,19 @@
+# Minimal image to serve the trained pipeline with FastAPI
+FROM python:3.11-slim
+WORKDIR /app
+
+# Pin runtime deps (match training!)
+RUN pip install --no-cache-dir \
+    fastapi==0.115.* \
+    "uvicorn[standard]==0.30.*" \
+    joblib==1.4.2 \
+    numpy==1.26.* \
+    pandas==2.2.* \
+    scikit-learn==1.6.1   # <= match your training version
+
+COPY app.py diabetes_model.pkl /app/
+EXPOSE 7860
+
+# IMPORTANT: bind to the port Spaces provides
+CMD ["sh", "-c", "uvicorn app:app --host 0.0.0.0 --port ${PORT:-7860} --workers 1 --timeout-keep-alive 120"]
+

app.py

@@ -0,0 +1,38 @@
+import joblib
+import pandas as pd
+from fastapi import FastAPI
+from pydantic import BaseModel
+
+pipe = joblib.load("diabetes_model.pkl")
+
+# Columns from Nb3 (Pima Indians Diabetes)
+FEATURE_COLUMNS = [
+    "pregnancies","glucose","blood_pressure","skin_thickness",
+    "insulin","bmi","diabetes_pedigree","age"
+]
+
+class Patient(BaseModel):
+    pregnancies: float
+    glucose: float
+    blood_pressure: float
+    skin_thickness: float
+    insulin: float
+    bmi: float
+    diabetes_pedigree: float
+    age: float
+
+app = FastAPI(title="Diabetes Risk API", version="1.0.0")
+
+@app.post("/predict")
+def predict(p: Patient):
+    # Build a 1-row DataFrame with correct column names
+    row = pd.DataFrame([p.model_dump()])[FEATURE_COLUMNS]
+    yhat = int(pipe.predict(row)[0])
+
+    proba = None
+    try:
+        proba = float(pipe.predict_proba(row)[0][1])
+    except Exception:
+        pass
+
+    return {"prediction": yhat, "positive_probability": proba}

diabetes_model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eed9a6e5726598c53d39688024bea69bf0fa227c62efb829d9e7490d2eaccd5c
+size 1729691