import pandas as pd
from sklearn.model_selection import train_test_split
import os
from sklearn.metrics import r2_score
import xgboost as xgb
import matplotlib.pyplot as plt
pd.set_option('display.width',None)

def normal(s1):
    high = s1.describe()['75%'] + 1.5*(s1.describe()['75%']-s1.describe()['25%'])
    low = s1.describe()['25%'] - 1.5 * (s1.describe()['75%'] - s1.describe()['25%'])
    return s1[(s1>=low)&(s1<=high)]

df = pd.read_csv('区县400v入模数据.csv',encoding='gbk',index_col='dtdate')
df.index = pd.to_datetime(df.index)
print(df.head())

list_org = []
list_fl = []
list_sc = []
for org_name in df['org_name'].drop_duplicates():
    data = df[df['org_name']==org_name]
    if org_name.strip()[-4:] != '供电公司':
        continue
    data = data.loc[normal(data['0.4kv及以下']).index]
    X = data.drop(columns=['city_name','org_name','0.4kv及以下'])
    x = X.loc['2022-1':'2023-7'][:-3]
    x_eval = X.loc['2023-7']
    y = data['0.4kv及以下'].loc['2022-1':'2023-7'][:-3]
    y_eval = data['0.4kv及以下'].loc['2023-7']
    # plt.plot(range(len(y)),y)
    # plt.show()

    x_train,x_test,y_train,y_test = train_test_split(x,y,test_size=0.3,random_state=42)

    model = xgb.XGBRegressor(max_depth=6,learning_rate=0.05,n_estimators=150)
    model.fit(x_train,y_train)

    pred = model.predict(x_test)
    # print(org_name)
    list_org.append(org_name)
    # print(r2_score(pred,y_test))
    list_sc.append(r2_score(pred,y_test))
    predict = model.predict(x_eval)
    result = pd.DataFrame({'real':y_eval,'pred':predict},index=y_eval.index)
    # print(result)
    # print((result['real'][-3:]-result['pred'][-3:]).sum()/result['real'].sum())
    list_fl.append((result['real'][-3:]-result['pred'][-3:]).sum()/result['real'].sum())

df = pd.DataFrame({'org':list_org,'sc':list_sc,'goal':list_fl})
print(df)
print(df['goal'].value_counts(bins=[-0.05,-0.01,-0.005,0, 0.005, 0.01, 0.02,0.05],sort=False))