import numpy as np
import pandas as pd
from statsmodels.stats.anova import anova_lm
from statsmodels.formula.api import ols
from scipy.stats import stats
from tools.algo.humidity import absolute_humidity
from tools.algo.interpolation import interpolate_value
from datetime import datetime
import plotly.express as px

if __name__ == "__main__":
    df_weather = pd.read_csv("/home/juni/PycharmProjects/failure_analysis/data/weather/202007010000_202308310000_f.csv")
    df_failure = pd.read_excel("/home/juni/PycharmProjects/failure_analysis/data/failure/대옹 작업내용.xlsx")
    times = df_failure["작업 시작 시간"]
    temp = []
    humidity = []

    for time in times:
        temp.append(interpolate_value(df_weather, timestamp=time, col_key="기온", k=6))
        humidity.append(interpolate_value(df_weather, timestamp=time, col_key="상대습도", k=6))

    absolute_humidity = absolute_humidity(humidity, temp)

    df_failure["기온"] = temp
    df_failure["상대습도"] = humidity
    df_failure["절대습도"] = absolute_humidity
    df_failure['불량 여부'] = np.where(df_failure['불량 여부'] == 'F', 0, df_failure['불량 여부'])
    df_failure['불량 여부'] = np.where(df_failure['불량 여부'] == 'T', 1, df_failure['불량 여부'])

    df_failure = df_failure.loc[:,['불량 여부', '상대습도', '기온', '절대습도']]

    F_statistic, pVal = stats.f_oneway(df_failure[df_failure["불량 여부"] == 0].loc[:,['상대습도', '기온', '절대습도']],
                                       df_failure[df_failure["불량 여부"] == 1].loc[:,['상대습도', '기온', '절대습도']])
    fig_df = pd.DataFrame(np.transpose([F_statistic, pVal]), index=['상대습도', '기온', '절대습도'], columns=["F_", "pVal"])
    fig = px.scatter(fig_df)
    fig.add_hline(y=0.05, line_dash="dash", line_color="red")
    fig.show()
    print(fig_df)