generate_data_set.py

import pandas as pd
from datetime import datetime, timedelta
from TestBridgeTemperatureDT.query_weather_data import query_weather_data
from TestBridgeTemperatureDT.biased_dummy_model import BiasedDummyModel
from TestBridgeTemperatureDT.dummy_model import DummyModel
from astral.sun import zenith, azimuth, elevation
from astral import LocationInfo
import matplotlib.pyplot as plt
import h5py


def generate_data_set(
    api_key_path: str,
    city: str,
    start_date: str,
    end_date: str,
    output_path: str,
    show: bool = False,
    true_c1: float = 1.0,
    true_c2: float = 0.01,
    true_c3: float = 0.01,
    true_c4: float = 0.01,
    true_c5: float = 0.01,
):
    weather_data = query_weather_data(api_key_path, city, start_date, end_date)

    city_info = LocationInfo(city, "Europe", "Berlin", 49.63, 8.38)
    sun_azimuths = []
    sun_zeniths = []
    sun_elevations = []
    for ref_time in weather_data["reference_time"]:
        # s = sun(city_info.observer, date=datetime.fromtimestamp(ref_time))
        sun_azimuths.append(azimuth(city_info.observer, datetime.fromtimestamp(ref_time)))
        sun_zeniths.append(zenith(city_info.observer, datetime.fromtimestamp(ref_time)))
        sun_elevations.append(elevation(city_info.observer, datetime.fromtimestamp(ref_time)))

    temperatures = [d["temp"] for d in weather_data["temperature"]]
    wind_speeds = [d["speed"] for d in weather_data["wind"]]
    cloud_coverages = weather_data["clouds"]
    reference_time = weather_data["reference_time"]

    # Create a pandas DataFrame with the data we want to plot
    data = pd.DataFrame(
        {
            "Temperature": temperatures,
            "Wind Speed": wind_speeds,
            "Cloud Coverage": cloud_coverages.to_dict().values(),  # somehow this is a pandas Series, not a list
            "Sun Azimuth": sun_azimuths,
            "Sun Zenith": sun_zeniths,
            "Sun Elevation": sun_elevations,
        },
        index=reference_time,
    )

    # Store data as hdf5 at output path
    output_name, output_format = output_path.split('.')
    input_data_path = output_name + '_input.' + output_format
    data.to_hdf(input_data_path, key="data", mode="w")

    # Create a plot of the data
    if show:
        for col in data.columns:
            plt.plot(data.index, data[col])
            plt.title(col)
            plt.xlabel("Time")
            plt.ylabel(col)
            plt.show()

    # Create a dummy model
    model = BiasedDummyModel(true_c1, true_c2, true_c3, true_c4, true_c5, 10, 10)
    model.calculate_grid(temperatures, wind_speeds, sun_zeniths, sun_azimuths, sun_elevations)
    computational_model = DummyModel(true_c1, true_c2, true_c3, true_c4, true_c5, 11, 11)
    computational_model.calculate_grid(temperatures, wind_speeds, sun_zeniths, sun_azimuths, sun_elevations)
    output_name, output_format = output_path.split('.')
    output_data_path = output_name + '_output.' + output_format
    with h5py.File(output_data_path, "w") as f:
        f.create_dataset("Model output", data=model.grid[5,0,:])
    model.plot_with_slider()
    error_model = DummyModel(true_c1, true_c2, true_c3, true_c4, true_c5, 11, 11)
    error_model.grid = computational_model.grid - model.grid
    error_model.t_size = computational_model.t_size
    # error_model.plot_with_slider(vmin=-1.0, vmax=1.0)

    plt.figure()
    plt.plot(model.grid[5,0,:],'.', label='Model')
    plt.plot(computational_model.grid[5,0,:],'.', label='Computational Model')
    plt.show()


    # return data


if __name__ == "__main__":
    output_path = "example.hdf5"
    api_key_path = "/home/darcones/Projects/API keys/owm.txt"
    city = "Worms"
    start_date = datetime.now().strftime("%Y-%m-%d")
    end_date = (datetime.now() - timedelta(days=5)).strftime("%Y-%m-%d")
    generate_data_set(api_key_path, city, start_date, end_date, output_path=output_path, show=True)
    # df.to_json('example_full.json')