src/predict.py

import os
from datetime import date, timedelta

import joblib
import pandas as pd
from dotenv import load_dotenv
from huggingface_hub import hf_hub_download, login

from src.data_api_calls import get_combined_data
from src.features_pipeline import create_features


def load_model(particle):
    load_dotenv()
    login(token=os.getenv("HUGGINGFACE_DOWNLOAD_TOKEN"))

    repo_id = f"elisaklunder/Utrecht-{particle}-Forecasting-Model"
    if particle == "O3":
        file_name = "O3_svr_model.pkl"
    elif particle == "NO2":
        file_name = "NO2_svr_model.pkl"

    model_path = hf_hub_download(repo_id=repo_id, filename=file_name)
    model = joblib.load(model_path)
    return model


def run_model(particle, data):
    input_data = create_features(data=data, target_particle=particle)
    model = load_model(particle)
    prediction = model.predict(input_data)
    target_scaler = joblib.load(f"scalers/target_scaler_{particle}.joblib")
    prediction = target_scaler.inverse_transform(prediction)
    return prediction


def get_data_and_predictions():
    PREDICTIONS_FILE = "predictions_history.csv"

    week_data = get_combined_data()

    o3_predictions = run_model("O3", data=week_data)
    no2_predictions = run_model("NO2", data=week_data)

    prediction_data = []
    for i in range(3):
        prediction_data.append(
            {
                "pollutant": "O3",
                "date_predicted": date.today(),
                "date": date.today() + timedelta(days=i + 1),
                "prediction_value": o3_predictions[0][i],
            }
        )
        prediction_data.append(
            {
                "pollutant": "NO2",
                "date_predicted": date.today(),
                "date": date.today() + timedelta(days=i + 1),
                "prediction_value": no2_predictions[0][i],
            }
        )

    predictions_df = pd.DataFrame(prediction_data)

    if os.path.exists(PREDICTIONS_FILE):
        existing_data = pd.read_csv(PREDICTIONS_FILE)
        combined_data = pd.concat([existing_data, predictions_df])
        combined_data = combined_data.drop_duplicates(
            subset=["pollutant", "date_predicted", "date"], keep="first"
        )
    else:
        combined_data = predictions_df

    combined_data.to_csv(PREDICTIONS_FILE, index=False)
    return week_data, o3_predictions, no2_predictions