app.py

from fastapi import FastAPI, Form
from fastapi.responses import HTMLResponse
from fastapi.templating import Jinja2Templates
from fastapi.requests import Request
import pickle
import pandas as pd
from typing import Optional

app = FastAPI()
templates = Jinja2Templates(directory="templates")

model = pickle.load(open("hiring_model.pkl", "rb"))


def normalize(value, cast=None):
    if value in (None, "", " "):
        return None
    return cast(value) if cast else value


def predict_job(job_data):
    import numpy as np
    
    # Replace None with np.nan to ensure numeric dtype
    clean_data = {k: (np.nan if v is None else v) for k, v in job_data.items()}
    df = pd.DataFrame([clean_data])
    
    # Force all columns to float64 to match training data
    df = df.astype('float64')
    
    # Predict
    probability = model.predict_proba(df)[0][1]
    prediction = "HIRED" if probability > 0.5 else "NOT_HIRED"
    return {
        'prediction': prediction,
        'hire_probability': f"{probability*100:.1f}",
        'confidence': 'High' if probability > 0.7 or probability < 0.3 else 'Medium'
    }


@app.get("/", response_class=HTMLResponse)
def home(request: Request):
    return templates.TemplateResponse("index.html", {"request": request})


@app.post("/predict", response_class=HTMLResponse)
def predict(
    request: Request,
    client_hire_rate: Optional[float] = Form(None),
    client_age_years: Optional[float] = Form(None),
    client_active_hires: Optional[int] = Form(None),
    client_avg_hourly_rate: Optional[float] = Form(None),
    client_total_reviews: Optional[int] = Form(None),
    client_total_hires: Optional[int] = Form(None),
    client_total_hours: Optional[int] = Form(None),
    client_total_spent: Optional[float] = Form(None),
    client_rating: Optional[float] = Form(None),
):
    job_data = {
        "client_hire_rate": normalize(client_hire_rate, float),
        "client_age_years": normalize(client_age_years, float),
        "client_active_hires": normalize(client_active_hires, int),
        "client_avg_hourly_rate": normalize(client_avg_hourly_rate, float),
        "client_total_reviews": normalize(client_total_reviews, int),
        "client_total_hires": normalize(client_total_hires, int),
        "client_total_hours": normalize(client_total_hours, int),
        "client_total_spent": normalize(client_total_spent, float),
        "client_rating": normalize(client_rating, float),
    }

    print(f"job data: {job_data}")

    result = predict_job(job_data)

    return templates.TemplateResponse(
        "index.html",
        {"request": request, "result": result}
    )