ARIMA模型识别、计算p、q值

#-*- coding: utf-8 -*-
#确定最佳p、d、q值
import pandas as pd

#参数初始化
discfile = '../data/discdata_processed.xls'

data = pd.read_excel(discfile, index_col = 'COLLECTTIME')
data = data.iloc[: len(data)-5] #不使用最后5个数据
xdata = data['CWXT_DB:184:D:\\']

from statsmodels.tsa.arima_model import ARIMA

#定阶
pmax = int(len(xdata)/10) #一般阶数不超过length/10
qmax = int(len(xdata)/10) #一般阶数不超过length/10
bic_matrix = [] #bic矩阵
for p in range(pmax+1):
  tmp = []
  for q in range(qmax+1):
    try: #存在部分报错，所以用try来跳过报错。
      tmp.append(ARIMA(xdata, (p,1,q)).fit().bic)
    except:
      tmp.append(None)
  bic_matrix.append(tmp)
print(bic_matrix)
#[[1275.6868239439104, 1273.190434524266, 1273.5749982328914, 1274.4669152438114, None], 
#[1276.7491283595593, 1271.8999324285992, None, None, None], 
#[1279.6942963992901, 1277.5553412371614, None, 1280.0924824267408, None],
# [1278.0659994468958, 1278.9885944429066, 1282.782534558853, 1285.943493708969, None],
# [1281.220790614283, 1282.6999920212124, 1286.2975191780365, 1290.1950373803218, None]]
bic_matrix = pd.DataFrame(bic_matrix) #从中可以找出最小值
print(bic_matrix)
#              0            1            2            3     4
# 0  1275.686824  1273.190435  1273.574998  1274.466915  None
# 1  1276.749128  1271.899932          NaN          NaN  None
# 2  1279.694296  1277.555341          NaN  1280.092482  None
# 3  1278.065999  1278.988594  1282.782535  1285.943494  None
# 4  1281.220791  1282.699992  1286.297519  1290.195037  None
print(bic_matrix.stack())
# 0  0    1275.69
#    1    1273.19
#    2    1273.57
#    3    1274.47
# 1  0    1276.75
#    1     1271.9
# 2  0    1279.69
#    1    1277.56
#    3    1280.09
# 3  0    1278.07
#    1    1278.99
#    2    1282.78
#    3    1285.94
# 4  0    1281.22
#    1     1282.7
#    2     1286.3
#    3     1290.2
p,q = bic_matrix.stack().astype('float64').idxmin() #先用stack展平，然后用idxmin找出最小值位置。
print(u'BIC最小的p值和q值为：%s、%s' %(p,q))