pytorch/浙江电压等级电量/区县分压/区县分压.py

import numpy as np
import pandas as pd
from prophet import Prophet
import math
import matplotlib.pyplot as plt
import os
from openpyxl import Workbook
pd.set_option('display.width',None)
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<=high)&(x>=low)


df = pd.read_csv(r'C:\Users\鸽子\Desktop\浙江各区县数据(2).csv')
df.columns = df.columns.map(lambda x:x.strip())
df.dropna(subset=['city_name','county_name'],inplace=True)
print(df.info())
print(df.columns)

print(dict(zip(df.columns,[(df[x]==0).sum()/len(df) for x in df.columns])))
yc_org_list = []
list_fl = []
list_org = []
list1 = []
for city in df['city_name'].drop_duplicates():
    wb = Workbook()
    wb.save(fr'C:\Users\鸽子\Desktop\11月区县分压预测\{city}.xlsx')

for org in df['county_name'].drop_duplicates():
    if org.strip()[-4:] != '供电公司':
        continue
    df_org = df[df['county_name']==org]
    city = df_org['city_name'].iloc[0]
    df_result = pd.DataFrame({})
    for level in df_org.columns[3:]:
        s1 = df_org[['pt_date',level]]
        s1.replace(0,np.NaN,inplace=True)
        s1.dropna(how='any',inplace=True)

        # 更改列名，更改为Prophet指定的列名ds和y
        dd = s1.rename(columns={'pt_date':'ds',level:'y'})
        dd['ds'] = dd['ds'].map(lambda x:x.strip())
        dd['ds'] = pd.to_datetime(dd['ds'])
        dd.drop_duplicates(inplace=True)

        # 划分数据，划分为训练集和验证集，预测的数据设置为未来4天
        df_train = dd[(dd['ds']>='2023-01-01')&(dd['ds']<='2023-11-30')]

        # df_train = df_train.loc[normal(df_train['y']).index]
        df_train['y'] = df_train['y'].where(normal(df_train['y']),other=np.nan).bfill()

        if df_train.shape[0] <= 90:
            yc_org_list.append(org)
            continue
        # df_test = dd[(dd['ds']>='2022-01-01')&(dd['ds']<='2023-07-31')][-3:]
        # 数据的变动会受到季节、周、天的影响，存在一定的规律性，因此我们将这三个参数设置为True
        model = Prophet(yearly_seasonality=True, weekly_seasonality=True, daily_seasonality=True)
        # 采用中国的假期模式，其余参数均保持默认
        model.add_country_holidays(country_name="CN")

        model.fit(df_train)
        # make_future_dataframe: 作用是告诉模型我们要预测多长时间，以及时间的周期是什么。生成一个时间戳
        future = model.make_future_dataframe(periods=4, freq='D')

        # 进行预测，返回预测的结果forecast
        forecast = model.predict(future)
        # forecast['additive_terms'] = forecast['weekly'] + forecast['yearly']；
        # 有：forecast['yhat']  = forecast['trend'] +  forecast['additive_terms'] 。
        # 因此：forecast['yhat']  = forecast['trend'] +forecast['weekly'] + forecast['yearly']。
        # 如果有节假日因素，那么就会有forecast['yhat']  = forecast['trend'] +forecast['weekly'] + forecast['yearly'] + forecast['holidays']。

        # 测试，把ds列，即data_series列设置为索引列
        # df_test = df_test.set_index('ds')

        # 把预测到的数据取出ds列，预测值列yhat，同样把ds列设置为索引列。
        forecast = forecast[['ds','yhat']].set_index('ds').sort_index(ascending=True).loc['2023-11']

        # 将预测列前25天替换为真实值
        forecast.loc['2023-11'][:25] = dd.set_index('ds').loc['2023-11'][:25]

        if len(forecast) < 334:
            list1.append(org)
        # join:按照索引进行连接，
        forecast.columns = [level]

        df_result = pd.concat([df_result,forecast],axis=1)

        # df_all = forecast.join(dd.set_index('ds')).dropna()
        # df_all['org_name'] = org
        # df_all['偏差率'] = (df_all['y'] - df_all['yhat'])/df_all['y']
        # df_all.rename(columns={'y':'真实值','yhat':'预测值'},inplace=True)
        # df_all = df_all[['org_name','真实值','预测值','偏差率']]

        list_org.append(org)

        # result = df_all.loc['2023-7']
        # result['goal'] = (result['真实值'] - result['预测值'])[-3:].sum()/result['真实值'].sum()
        # list_fl.append((result['真实值'] - result['预测值'])[-3:].sum()/result['真实值'].sum())

    with pd.ExcelWriter(fr'C:\Users\鸽子\Desktop\11月区县分压预测\{city}.xlsx',mode='a',engine='openpyxl',if_sheet_exists='replace') as writer:
        df_result.to_excel(writer,sheet_name=f'{org}')

print(yc_org_list)
df = pd.DataFrame({'org':list_org,'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))

    # # 创建一个ExcelWriter对象
    # with pd.ExcelWriter(r'C:\Users\鸽子\Desktop\output.xlsx',mode='a',if_sheet_exists='replace') as writer:
    #     # 将不同的子文件写入同一个Excel文件的不同工作表
    #     df_all.to_excel(writer, sheet_name=f'Sheet{i+1}')

# df_all.plot()
# # 设置左上角小标
# plt.legend(['true', 'yhat'])
# plt.show()