mvlick/atisfaction-prediction-api

1from fastapi import FastAPI2from pydantic import BaseModel3import pandas as pd4import joblib5 6# ===============================7# ⚙️ Configuration8# ===============================9MODEL_PATH = "model_best.joblib"  # ou model.pkl selon ton fichier10model = joblib.load(MODEL_PATH)11 12app = FastAPI(13    title="Satisfaction Prediction API",14    description="API de prédiction de satisfaction apprenants",15    version="1.0.0"16)17 18# ===============================19# 📦 Schémas d’entrée20# ===============================21class Diploma(BaseModel):22    level: str = ""23    title: str = ""24 25class Experience(BaseModel):26    title: str = ""27    description: str = ""28 29class PastCourse(BaseModel):30    title: str = ""31    description: str = ""32    numberOfStars: float | None = None33 34class Professor(BaseModel):35    fistname: str = ""36    lastname: str = ""37    city: str = ""38    description: str = ""39    diplomas: list[Diploma] = []40    experiences: list[Experience] = []41    pastCourses: list[PastCourse] = []42 43class Course(BaseModel):44    title: str = ""45    description: str = ""46 47class InputData(BaseModel):48    professor: Professor49    course: Course50 51# ===============================52# 🧩 Prétraitement53# ===============================54def adapt_input(body: dict) -> dict:55    prof = body.get("professor", {})56    course = body.get("course", {})57 58    diplomas = prof.get("diplomas", [])59    exps = prof.get("experiences", [])60    past_courses = prof.get("pastCourses", [])61 62    # Diplômes63    levels = [d.get("level", "").lower() for d in diplomas]64    has_master = int(any("master" in lvl for lvl in levels))65    has_licence = int(any("licence" in lvl or "bachelor" in lvl for lvl in levels))66    has_doctorat = int(any("doctor" in lvl or "phd" in lvl for lvl in levels))67    has_certif = int(any("certif" in lvl for lvl in levels))68    has_secondary = int(any("second" in lvl for lvl in levels))69 70    # Expériences71    all_exp_text = " ".join(72        f"{e.get('title', '')} {e.get('description', '')}".lower()73        for e in exps74    )75    has_teaching_exp = int(any(k in all_exp_text for k in ["prof", "enseign", "formateur"]))76    has_industry_exp = int(any(k in all_exp_text for k in ["dev", "engineer", "ingénieur", "consult", "industrie"]))77    has_management_exp = int(any(k in all_exp_text for k in ["chef", "lead", "manager", "responsable"]))78    has_academic_exp = int(any(k in all_exp_text for k in ["recherche", "thèse", "doctorat", "universit"]))79 80    # Moyenne et nombre de cours passés81    past_ratings = [c.get("numberOfStars") for c in past_courses if c.get("numberOfStars") is not None]82    teacher_mean_excl = float(sum(past_ratings) / len(past_ratings)) if past_ratings else 083    teacher_course_count = len(past_courses)84 85    # Infos générales86    desc_len_words = len(prof.get("description", "").split())87    n_diplomas = len(diplomas)88    n_experiences = len(exps)89 90    # Nouveau cours91    course_title = course.get("title", "").strip().lower()92    course_description = course.get("description", "").strip().lower()93    full_course_text = f"{course_title} {course_description}".strip()94 95    # ✅ Données prêtes pour le modèle96    return {97        "course_title": full_course_text,98        "desc_len_words": desc_len_words,99        "n_diplomas": n_diplomas,100        "n_experiences": n_experiences,101        "teacher_mean_excl": teacher_mean_excl,102        "teacher_course_count": teacher_course_count,103        "has_master": has_master,104        "has_licence": has_licence,105        "has_doctorat": has_doctorat,106        "has_certif": has_certif,107        "has_secondary": has_secondary,108        "has_teaching_exp": has_teaching_exp,109        "has_industry_exp": has_industry_exp,110        "has_management_exp": has_management_exp,111        "has_academic_exp": has_academic_exp112    }113 114# ===============================115# 🚀 Endpoint principal116# ===============================117@app.post("/api/predict")118def predict(data: InputData):119    body = data.dict()120    features = adapt_input(body)121    df = pd.DataFrame([features])122 123    prediction = model.predict(df)[0]124    prediction = round(float(prediction), 2)125 126    return {"gradeAverage": prediction}127 128# ===============================129# 🧪 Test rapide130# ===============================131@app.get("/")132def root():133    return {134        "message": "Bienvenue sur l’API de prédiction de satisfaction 🎯",135        "usage": "POST /api/predict avec le JSON d’un professeur et d’un cours"136    }