Singular-Bean/sequence-sim

1from fastapi import FastAPI, Request, Response2from fastapi.middleware.cors import CORSMiddleware3import joblib4import numpy as np5import pandas as pd6from xgboost import XGBClassifier7 8app = FastAPI()9 10app.add_middleware(11    CORSMiddleware,12    allow_origins=["*"],13    allow_credentials=True,14    allow_methods=["*"],15    allow_headers=["*"],16)17 18choiceModel = joblib.load('models/choice_predictor.pkl')19passModel = joblib.load('models/pass_predictor.pkl')20moveModel = joblib.load('models/movement_predictor.pkl')21xModel = joblib.load('models/x_predictor.pkl')22yModel = joblib.load('models/y_predictor.pkl')23 24def convertCoordsTo(x=None, y=None):25    if x is not None:26        return 3+(x*1.05)27    elif y is not None:28        return 3+(y*0.68)29 30def convertCoordsFro(x=None, y=None):31    if x is not None:32        return (x-3)/1.0533    elif y is not None:34        return (y-3)/0.6835 36def predictNextPass(df, xModel, yModel, temperature=0.75):37    xDist = xModel.pred_dist(df[['xStart', 'yStart']])38    mu_x = xDist.loc[0]39    sigma_x = xDist.scale[0]40    simulated_x = np.random.normal(mu_x, sigma_x * temperature)41    while simulated_x < 0 or simulated_x > 100:42        simulated_x = np.random.normal(mu_x, sigma_x * temperature)43    df['xEnd'] = simulated_x44    yDist = yModel.pred_dist(df[['xStart', 'yStart', 'xEnd']])45    mu_y = yDist.loc[0]46    sigma_y = yDist.scale[0]47    simulated_y = np.random.normal(mu_y, sigma_y * temperature)48    while simulated_y < 0 or simulated_y > 100:49        simulated_y = np.random.normal(mu_y, sigma_y * temperature)50    return simulated_x, simulated_y51 52@app.get("/")53@app.get("/{cache_bust}")54async def hello(response: Response, cache_bust: str | None = None):55    response.headers["Cache-Control"] = "no-store, no-cache, must-revalidate, max-age=0"56    response.headers["Pragma"] = "no-cache"57    response.headers["Expires"] = "0"58    return {"success": True}59 60@app.post("/")61async def calc_path(request: Request):62    payload = await request.json()63    print(payload)64 65    inputX = convertCoordsFro(x=payload['end']['x'])66    inputY = convertCoordsFro(y=payload['end']['y'])67 68    def nextChoice(x, y):69        y_pred = choiceModel.predict_proba(pd.DataFrame({'1': {'x': float(x), 'y': float(y)}}).T)70        raw_probs = y_pred[0]71 72        probabilities = raw_probs / raw_probs.sum()73 74        options = ['Pass', 'Shot', 'Dribble']75 76        choice = np.random.choice(options, p=probabilities)77        return choice78 79    def passDestination(x, y):80        predX, predY = predictNextPass(pd.DataFrame({'1': {'xStart': float(x), 'yStart': float(y)}}).T, xModel, yModel)81        return predX, predY82 83    def dribbleDestination(x, y):84        y_pred = moveModel.predict(pd.DataFrame({'1': {'xStart': float(x), 'yStart': float(y)}}).T)85        return y_pred[0]86 87    sequence = []88    switch = True89    while switch:90        choice = nextChoice(inputX, inputY)91        if choice == 'Shot':92            sequence.append({"type": "shot", "x": 108, "y": 37})93            switch = False94        elif choice == 'Pass':95            inputX, inputY = passDestination(inputX, inputY)96            sequence.append({"type": "pass", "x": convertCoordsTo(x=inputX), "y": convertCoordsTo(y=inputY)})97        elif choice == 'Dribble':98            inputX, inputY = dribbleDestination(inputX, inputY)99            sequence.append({"type": "dribble", "x": convertCoordsTo(x=inputX), "y": convertCoordsTo(y=inputY)})100    return sequence101