Create main.py

main.py

@@ -0,0 +1,71 @@
+from fastapi import FastAPI
+from fastapi.middleware.cors import CORSMiddleware
+from pydantic import BaseModel, Field, validator
+from typing import List, Dict, Any
+import os
+
+from inference import load_model, predict_one, predict_batch, repo_snapshot
+
+HF_REPO_ID = os.getenv("HF_REPO_ID", "ethnmcl/test-score-predictor-xgb")
+
+app = FastAPI(title="Test Score Predictor API",
+              version="1.0.0",
+              description="FastAPI wrapper for ethnmcl/test-score-predictor-xgb")
+
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"], allow_credentials=True,
+    allow_methods=["*"], allow_headers=["*"],
+)
+
+# Load model at startup (downloads snapshot if not already present)
+@app.on_event("startup")
+def _startup():
+    repo_snapshot(HF_REPO_ID)  # ensures files exist locally
+    load_model()               # loads artifacts into process
+
+
+class Record(BaseModel):
+    Subject: str = Field(..., examples=["Mathematics"])
+    Current_Grade: int = Field(..., ge=60, le=98)
+    Max_Test_Percentage: int = Field(..., ge=65, le=100)
+    Days_Preparing: int = Field(..., ge=1, le=14)
+    Hours_Studied: int = Field(..., ge=2, le=50)
+    Study_Session_Average: float = Field(..., ge=0.1, le=10.0)
+    Avg_Previous_Tests: int = Field(..., ge=55, le=95)
+    Test_Difficulty: str = Field(..., examples=["Easy (20)", "Medium (30)", "Hard (50)"])
+
+    @validator("Study_Session_Average", always=True)
+    def recompute_session_avg(cls, v, values):
+        # Keep dataset contract: Hours / Days, rounded to 1 decimal
+        if "Hours_Studied" in values and "Days_Preparing" in values:
+            h = values["Hours_Studied"]; d = values["Days_Preparing"]
+            return round(h / d, 1)
+        return v
+
+
+class PredictRequest(BaseModel):
+    data: List[Record]
+
+
+@app.get("/health")
+def health() -> Dict[str, Any]:
+    return {"status": "ok", "repo": HF_REPO_ID}
+
+
+@app.get("/model-info")
+def model_info() -> Dict[str, Any]:
+    return {"repo": HF_REPO_ID, "files": ["preprocessor.joblib", "weights.npy", "xgb_model.json", "schema.json"]}
+
+
+@app.post("/predict")
+def predict(req: Record) -> Dict[str, Any]:
+    score = predict_one(req.dict())
+    return {"predicted_score": float(score)}
+
+
+@app.post("/predict-batch")
+def predict_many(req: PredictRequest) -> Dict[str, Any]:
+    records = [r.dict() for r in req.data]
+    scores = predict_batch(records)
+    return {"predicted_scores": [float(s) for s in scores], "count": len(scores)}