predictV1/routes/predict_with_players.py

from catboost import CatBoostClassifier
import pandas as pd
import numpy as np
from typing import Dict, Any

# Загрузка модели с игроками
modelWithPlayers = CatBoostClassifier()
modelWithPlayers.load_model("artifacts/model_with_players.cbm")

# Загрузка порядка фич
def load_feature_order(path: str) -> list:
    fo = pd.read_csv(path)
    first_col = fo.columns[0]
    return fo[first_col].tolist()

FEATURE_ORDER_WITH_PLAYERS = load_feature_order("artifacts/feature_order_with_players.csv")

def build_player_features(payload: Dict[str, Any]) -> pd.DataFrame:
    """
    Создаёт бинарные признаки для модели с игроками.

    Признаки:
    - radiant_p{player_id}_h{hero_id}_pos{position}
    - radiant_p{player_id}_h{hero_id}
    - radiant_p{player_id}_pos{position}
    (аналогично для dire)
    """
    features = {}

    # Инициализируем все признаки нулями
    for feat in FEATURE_ORDER_WITH_PLAYERS:
        features[feat] = 0

    # Radiant: игроки + герои + позиции
    for i in range(1, 6):
        hero_id = int(payload.get(f"r_h{i}", -1))
        player_id = int(payload.get(f"r_p{i}", -1))
        position = int(payload.get(f"rp_h{i}", -1))

        # Признак: игрок + герой + позиция
        if player_id > 0 and hero_id >= 0 and position >= 0:
            feature_name = f"radiant_p{player_id}_h{hero_id}_pos{position}"
            if feature_name in features:
                features[feature_name] = 1

        # Признак: только игрок + герой
        if player_id > 0 and hero_id >= 0:
            feature_name = f"radiant_p{player_id}_h{hero_id}"
            if feature_name in features:
                features[feature_name] = 1

        # Признак: только игрок + позиция
        if player_id > 0 and position >= 0:
            feature_name = f"radiant_p{player_id}_pos{position}"
            if feature_name in features:
                features[feature_name] = 1

    # Dire: игроки + герои + позиции
    for i in range(1, 6):
        hero_id = int(payload.get(f"d_h{i}", -1))
        player_id = int(payload.get(f"d_p{i}", -1))
        position = int(payload.get(f"dp_h{i}", -1))

        # Признак: игрок + герой + позиция
        if player_id > 0 and hero_id >= 0 and position >= 0:
            feature_name = f"dire_p{player_id}_h{hero_id}_pos{position}"
            if feature_name in features:
                features[feature_name] = 1

        # Признак: только игрок + герой
        if player_id > 0 and hero_id >= 0:
            feature_name = f"dire_p{player_id}_h{hero_id}"
            if feature_name in features:
                features[feature_name] = 1

        # Признак: только игрок + позиция
        if player_id > 0 and position >= 0:
            feature_name = f"dire_p{player_id}_pos{position}"
            if feature_name in features:
                features[feature_name] = 1

    # Создаём DataFrame с одной строкой в правильном порядке
    df = pd.DataFrame([features], columns=FEATURE_ORDER_WITH_PLAYERS)

    return df

def predict_with_players(payload: Dict[str, Any]) -> Dict[str, float]:
    """
    Делает предсказание с использованием модели с игроками.

    Возвращает:
    {
        "radiant_win": вероятность победы Radiant (0-100),
        "dire_win": вероятность победы Dire (0-100)
    }
    """
    # Проверяем, есть ли хотя бы один игрок в payload
    has_players = False
    for i in range(1, 6):
        if payload.get(f"r_p{i}", -1) > 0 or payload.get(f"d_p{i}", -1) > 0:
            has_players = True
            break

    # Если нет игроков, возвращаем 50/50
    if not has_players:
        return {
            "radiant_win": 50,
            "dire_win": 50
        }

    # Создаём признаки
    X = build_player_features(payload)

    # Предсказание
    proba = modelWithPlayers.predict_proba(X)[0, 1]

    radiant_win = round(float(np.clip(proba * 100.0, 0.0, 100.0)))
    dire_win = 100 - radiant_win

    return {
        "radiant_win": radiant_win,
        "dire_win": dire_win
    }