gistfile1.txt

#!/usr/bin/env python3
from fbprophet import Prophet
from glob import glob
from pandas import Series,DataFrame
from os.path import join,dirname,basename
from datetime import date
from collections import defaultdict
import pandas as pd
import csv
import re

RE_NUM = re.compile("\\d+")

def next_quarter(numLi):
    m = numLi[1]+3
    y = numLi[0]
    if m > 12:
        m = m % 12
        y += 1
    return (y,m,1)

DEPTH = 3 * 4

# 日期 = "2016-12-31  2017-09-30  2017-12-31  2018-09-30  2018-12-31  2019-03-31  2019-06-30  2019-09-30  2019-12-31  2020-03-31".split()
# 数据 = dict(
# 佣金收入 = "132.01  327.29  524.43  632.04  708.20  635.48  677.21  624.44  732.67  1427.30",
# 金融服务收入 = "3.27  32.28  82.89  157.42  171.93  208.61  191.14  189.89  203.04  163.62",
# 利息收入 = "1.08  5.29  51.88  259.84  505.59  833.21  477.30",
# 其他收入 = "1.62  2.45  1.52  10.98  32.14  64.03  162.08  299.10  228.09  250.56"
# )
#
# 预测 = defaultdict(list)

def extend(ds, li):
    df = DataFrame({
        "ds":list(map(lambda x:str(date(*x)),ds)),
        "y":li
    })
    m = Prophet(
        weekly_seasonality=False,
        n_changepoints=len(li)//12,
        daily_seasonality=False,
        seasonality_prior_scale=.5,
        changepoint_range=1
    )
    m.fit(df)
    future = m.make_future_dataframe(periods=DEPTH,freq="Q",include_history=False)
    forecast = m.predict(future)
    future.tail()
    # r = forecast[['ds','yhat', 'yhat_lower', 'yhat_upper']]
    # print(list(forecast.yhat))
    return list(forecast.yhat)

#
# # fig1 = m.plot(forecast)
# # fig2 = m.plot_components(forecast)
#
# # from statsmodels.tsa.api import Holt
# # import numpy as np
# # import matplotlib.pyplot as plt
# #
# # fit = Holt(np.asarray(income)).fit(smoothing_level=1.406, smoothing_slope=0.2)
# #
# # for i in fit.forecast(4):
# #     print(round(i,2))
#
# # plt.figure()
# # plt.plot(income[:-1]+list(fit.forecast(5)))
# # plt.show()

_DIR = dirname(__file__)
def csv_future(csvname):
    with open(join(_DIR,"csv",csvname+'.csv'),encoding="utf-8-sig") as f:
        with open(join(_DIR,"future",csvname+'.csv'),"w", encoding="utf-8-sig") as out:
            out = csv.writer(out)
            f = csv.reader(f)
            dateLi = []
            for pos,i in enumerate(f):
                if pos:
                    key = i[0].strip()
                    # if key!="经纪佣金":
                    #     continue
                    # if key!="薪酬及福利费用":
                    #     continue
                    if not key:
                        continue
                    ds_li = []
                    val_li = []
                    row_li = []
                    org_li = []
                    for d,val in zip(dateLi, i[1:]):
                        val = val.strip()
                        if val:
                            val = val.replace(',','')
                            val = float(val)
                            org_li.append(val)
                            month = d[1]
                            quarter = month//3
                            if not ds_li or (ds_li[-1][0] != d[0]):
                                val = val/quarter
                            else:
                                val = (val-org_li[-2]) /((d[1] - ds_li[-1][1])//3)
                            row_li.append(str(round(val,4)))
                            ds_li.append(d)
                            val_li.append(val)
                        else:
                            row_li.append("")
                    # print(val_li)
                    # print(ds_li)
                    out.writerow(
                        [key]+row_li+list(map(lambda x:str(round(x,4)),extend(ds_li,val_li)))
                    )
                else:
                    for d in i[1:]:
                        d = d.strip()
                        if not d:
                            break
                        d = tuple(map(int,RE_NUM.findall(d)))
                        dateLi.append(d)

                    li = dateLi[:]
                    ymd = dateLi[-1]
                    for i in range(DEPTH):
                        ymd = next_quarter(ymd)
                        li.append(ymd)

                    out.writerow([""]+list(
                            map(lambda x:str(date(*x))[:7],li)
                        )
                    )
if __name__ == "__main__":
    for i in glob(join(_DIR,"csv/*.csv")):
        i = basename(i)[:-4]
        print(i)
        csv_future(i)