import pandas as pd
import matplotlib.pyplot as plt
import xgboost as xgb
from sklearn.model_selection import train_test_split
from sklearn.metrics import r2_score
def normal(x):
    high = x.describe()['75%'] + 1.5*(x.describe()['75%']-x.describe()['25%'])
    low = x.describe()['25%'] - 1.5*(x.describe()['75%']-x.describe()['25%'])
    return x[(x<=high)&(x>=low)]

def season(x):
    if str(x)[5:7] in ('04', '05', '06', '11'):
        return 0
    elif str(x)[5:7] in ('01', '02', '03', '09', '10', '12'):
        return 1
    else:
        return 2

df = pd.read_excel('../400v入模数据/衢州.xlsx',index_col='stat_date')
df.index = pd.to_datetime(df.index)


x_train = df.loc['2021-7':'2023-7'][:-3].drop(columns='0.4kv及以下')
y_train = df.loc['2021-7':'2023-7'][:-3]['0.4kv及以下']
x_eval = df.loc['2023-7'].drop(columns='0.4kv及以下')
y_eval = df.loc['2023-7']['0.4kv及以下']

x_train,x_test,y_train,y_test = train_test_split(x_train,y_train,test_size=0.2,random_state=142)
model = xgb.XGBRegressor(max_depth=6,learning_rate=0.05,n_estimators=150)
model.fit(x_train,y_train)
y_pred = model.predict(x_test)
print(r2_score(y_test,y_pred))

predict = model.predict(x_eval)
result = pd.DataFrame({'eval':y_eval,'pred':predict},index=y_eval.index)
print(result)
print((result['eval'][-3:].sum()-result['pred'][-3:].sum())/result['eval'].sum())