def get_d():
data = index2period()
P = acorr_ljungbox(data)[1]
d_data = data
for i in range(1, 5):
d_data = d_data.diff().dropna()
P = acorr_ljungbox(d_data)[1]
if len(P[P<0.05]) / len(P) >= 0.5:
d = i
break
return d, d_data
def _setup_autocorrelation(self, residuals):
lags = range(1, self.AUTOCORR_MAX_LAG + 1)
# p_value is a probability of no autocorrelation present (separate value for each lag)
_, p_value = acorr_ljungbox(residuals, lags=lags)
self.autocorrelation = p_value <= self.AUTOCORR_SIGNIFICANCE_LEVEL
def _modlSignf(self, resid):
qljungbox, pval, qboxpierce, pvalbp = acorr_ljungbox(
resid, boxpierce=True) #只有当参数boxpierce=True时, 才会输出Q统计量.
print("modlSinf =")
print(qljungbox)
print(pval)
return min(pval) > 0.05
def pval_test():
data_1 = sequence_chart()
qljungbox, pval, qboxpierce, pvalbp = acorr_ljungbox(data_1,
boxpierce=True)
plt.plot(range(0, len(pval)), pval)
plt.plot(range(0, len(pvalbp)), pvalbp)
plt.legend()
plt.show()
def randomness(self, data=None):
"""
随机性检测
默认情况下, acorr_ljungbox只计算LB统计量, 只有当参数boxpierce=True时, 才会输出Q统计量.
一般如果统计量的P值小于0.05时，则可以拒绝原假设，认为该序列为非白噪声序列，跟Q统计量差不多。
:return:
"""
if data is None:
data = self.data
lbvalue, pval = acorr_ljungbox(data.dropna(), lags=True)
return (True, pval[0]) if pval[0] > 0.05 else (False, pval[0])
def arima():
series_ch = pd.read_csv(
"http://labfile.oss.aliyuncs.com/courses/1176/agriculture.csv",
index_col=0)
series_ch.plot(figsize=(9, 6))
fig, axes = plt.subplots(ncols=3, nrows=1, figsize=(15, 3))
diff_ch = series_ch.diff().dropna()
axes[0].plot(diff_ch)
autocorrelation_plot(diff_ch, ax=axes[1])
axes[2].plot(acorr_ljungbox(diff_ch)[1])
fig, axes = plt.subplots(ncols=3, nrows=1, figsize=(15, 3))
diff_ch1 = series_ch.diff(periods=2).dropna()
axes[0].plot(diff_ch1)
autocorrelation_plot(diff_ch1, ax=axes[1])
axes[2].plot(acorr_ljungbox(diff_ch1)[1])
fig, axes = plt.subplots(ncols=3, nrows=1, figsize=(15, 3))
diff_ch2 = series_ch.diff().diff().dropna()
axes[0].plot(diff_ch2)
autocorrelation_plot(diff_ch2, ax=axes[1])
axes[2].plot(acorr_ljungbox(diff_ch2)[1])
fig, axes = plt.subplots(ncols=3, nrows=1, figsize=(15, 3))
diff_ch3 = series_ch.diff().diff().diff().dropna()
axes[0].plot(diff_ch3)
autocorrelation_plot(diff_ch3, ax=axes[1])
axes[2].plot(acorr_ljungbox(diff_ch3)[1])
d = 1
p, q = arma_order_select_ic(diff_ch, ic='aic')['aic_min_order']
print('p,d,q', p, d, q)
return p, d, q
def draw_picture():
parameter_type = ['W01', '060', 'W02', '101', 'W07']
wdp_mode = {
'W01': '水温',
'060': '氨氮',
'W02': '溶解氧',
'101': '总磷',
'W07': '高锰酸盐'
}
data_set = {parameter: get_data(parameter) for parameter in parameter_type}
for key, value in data_set.items():
print(wdp_mode[key], acorr_ljungbox(value['data_value'], lags=6))
mpl.rcParams['font.sans-serif'] = ['SimHei'] #正常显示中文
mpl.rcParams['axes.unicode_minus'] = False
# plt.title(wdp_mode[key]) # 显示图标题
# plt.show()
return 0
def beforehand_test(df):
"""单变量检验"""
from scipy import stats
from statsmodels.sandbox.stats.diagnostic import acorr_ljungbox, het_arch
def adf_sets(xs):
adfs = {}
for c in xs:
adf = adfTest(xs[c])
adfs[c] = adf
return adfs
def statistic(df,l_ar_order=[1,15]):
cols = df.columns
from scipy import stats
skew = stats.skew(df)
kuro = stats.kurtosis(df)
ar_tvalues = {}
for _order in l_ar_order:
ar_tvalues["AR(%s)"%_order] = [sm.tsa.AR(df[c]).fit(_order).tvalues[-1] for c in df]
ar_tvalues = pd.DataFrame(ar_tvalues,index=cols).T
desb2 = pd.DataFrame(np.vstack((skew, kuro)), columns=cols, index=['skew', "kuro"])
desb = pd.concat([df.describe(),desb2,ar_tvalues],axis=0)
return desb
accor = {}
for i, c in enumerate(df):
s = df[c]
accor[c] = pd.Series([het_arch(s)[-1], \
(np.hsplit(np.array(acorr_ljungbox(s, 15)),15)[-1][-1])[0],
stats.jarque_bera(s)[1]],
index=['ARCH', 'LBQ', 'JB'])
return pd.concat([statistic(df),\
pd.DataFrame(adf_sets(df)).loc[['p']].rename({"p":"ADF(p)"}),\
pd.DataFrame(accor)])
def boxpierce_test():
# fig,ax = plt.subplots(2,2)
data_1 = sequence_chart()
qljungbox, pval, qboxpierce, pvalbp = acorr_ljungbox(data_1,
boxpierce=True)
print(len(pval))
# for i in range(len(pval)):
# print("true data:",qljungbox[i], pval[i], qboxpierce[i], pvalbp[i])
plt.plot(range(0, len(qljungbox)), qljungbox)
plt.plot(range(0, len(qboxpierce)), qboxpierce)
plt.legend()
# ax[0, 0].plot(range(0,len(pval)), qljungbox, label = "ql")
# ax[0, 0].plot(range(0,len(pval)), qboxpierce, label = "qb")
#
# ax[0,1].plot(range(0,len(pval)), label="pval",)
# ax[0,1].plot(range(0,len(pvalbp)), label="pvalbp")
# ax[0,0].legend()
# ax[0,1].legend()
plt.show()
def lb(x):
s,p = acorr_ljungbox(x, lags=4)
return np.r_[s, p]
def lb(x):
s, p = acorr_ljungbox(x, lags=4, return_df=True)
return np.r_[s, p]
def lb1(x):
s, p = acorr_ljungbox(x, lags=1, return_df=True)
return s[0], p[0]
def lb4(x):
s, p = acorr_ljungbox(x, lags=4, return_df=True)
return s[-1], p[-1]
def lb1(x):
s,p = acorr_ljungbox(x, lags=1)
return s[0], p[0]
def lb4(x):
s,p = acorr_ljungbox(x, lags=4)
return s[-1], p[-1]
print(u'原始序列的ADF检验结果为：', ADF(data[u'销量']))
# 返回值依次为adf(原始序列的单位根)、pvalue、usedlag、nobs、critical values、icbest、regresults、resstore
#p值0.99显著大于0.05，该序列为非平稳序列
#进行差分
D_data = data.diff().dropna() #diff沿着指定轴计算第N维的离散差值，即矩阵后一个元素减去前一个元素
D_data.columns = [u'销量差分']
D_data.plot() #差分后的时序图
plot_acf(D_data) #自相关图
plot_pacf(D_data) #偏自相关图
# plt.show()
print(u'差分序列的ADF检验结果为：', ADF(D_data[u'销量差分']))
#一阶差分后时序图在均值附近波动，自相关图有很强的短期相关性，p值0.02小于0.05，说明序列是平稳序列
#对一阶差分后的序列做白噪声检验
print(u'差分序列的白噪声检验结果为：', acorr_ljungbox(D_data, lags=1))
#输出为stat、p值，0.00077远小于0.05，所以是平稳白噪声序列
data[u'销量'] = data[u'销量'].astype(float)
#对平稳白噪声序列拟合ARMA模型
#1、先定级确定p,q
pmax = int(len(D_data) / 10) #一般阶数不超过length/10
qmax = int(len(D_data) / 10)
bic_matrix = [] #bic矩阵
data.dropna(inplace=True)
import warnings
warnings.filterwarnings('error')
for p in range(pmax + 1):
tmp = []
def lb1(x):
s, p = acorr_ljungbox(x, lags=1)
return s[0], p[0]
def findSimModel(ts):
'''
寻找序列的简单模型，如果不存在则返回空
:param ts: 时间序列
:return: 是否是简单模型，拟合之后的模型
'''
#对序列差分得到平稳序列
stationSeries = difStationary(ts)
exitSimModel = False
isSimModel = True
#查分的阶数
d = list(stationSeries.keys())[0]
p,q = getpq(stationSeries[d])
if len(p) > 8 or max(p) >=12:
isSimModel = False
print('自相关系数超过两倍标准差的阶数有：{}'.format(p))
if len(q) > 8 or max(p) >=12:
isSimModel = False
print('偏自相关系数超过两倍标准差的阶数：{}'.format(q))
#确定该序列的p，q
if isSimModel:
ARp = max(p)
MAq = max(q)
else:
res = sm.tsa.arma_order_select_ic(y, ic=['aic'], trend='nc')
ARp = int(res.aic_min_order[0])
MAq = int(res.aic_min_order[1])
print('AIC MIN ORDER:{}'.format(res.aic_min_order))
model = SARIMAX(stationSeries[d],
order=(1,d,1),
seasonal_order=(0, 0, 0, 12),
enforce_stationarity=False,
enforce_invertibility=False)
results = model.fit()
p_values = results.pvalues
t_values = results.tvalues
resid = results.resid
qljungbox, pval, qboxpierce, pvalbp=acorr_ljungbox(resid, boxpierce=True) #只有当参数boxpierce=True时, 才会输出Q统计量.
print("qlb=")
print(qljungbox[0:60])
print("pval=")
print(pval[0:60])
print("t_values = ")
print(t_values)
print("p_value = ")
print(p_values)
paraSign = paraSignf(p_values)
modlSign = modlSignf(pval)
print(paraSign)
print(modlSign)
if paraSign and modlSign:
exitSimModel = True
if exitSimModel:
print("存在合适的拟合模型！")
return isSimModel,model
else:
print("没有合适的拟合模型！")
return isSimModel,None
def isWN(ts):
qljungbox, pval, qboxpierce, pvalbp=acorr_ljungbox(ts, boxpierce=True) #只有当参数boxpierce=True时, 才会输出Q统计量.
print(qljungbox)
print(pval)
return not(modlSignf(pval))
line = f.readline()
pre_565 = []
aft_565 = []
for i in list_pre:
pre_565.append(i[565])
aft_565.append(wt(i, 'db3', 4, 1, 4)[565])
import matplotlib.pyplot as plt
x = np.arange(0,1000)
y1 = pre_565
y2 = aft_565
qljungbox, pval, qboxpierce, pvalbp = acorr_ljungbox(y1, boxpierce=True)
plt.plot(range(0, len(pval)), pval)
plt.plot(range(0, len(pvalbp)), pvalbp)
plt.legend()
plt.show()
plt.plot(x,y1)
plt.plot(x,y2)
plt.show()
def lb4(x):
s, p = acorr_ljungbox(x, lags=4)
return s[-1], p[-1]
#!/usr/bin/env python3
import pandas as pd
import statsmodels.tsa.stattools as ts
from statsmodels.sandbox.stats.diagnostic import acorr_ljungbox
if __name__ == "__main__":
allData = pd.read_csv('MonthlyWeather.txt', header=None, sep=',')
data = allData.iloc[:, 0]
print(ts.adfuller(data, autolag='AIC'))
p_value = acorr_ljungbox(data, lags=[6, 12])
print(p_value)
def lb(x):
s,p = acorr_ljungbox(x, lags=4)
return np.r_[s, p]
from statsmodels.tsa.arima_model import ARMA
from statsmodels.sandbox.stats.diagnostic import acorr_ljungbox
import matplotlib.pyplot as plt
if __name__ == "__main__":
allData = pd.read_csv('MonthlyWeather.txt', header=None, sep=',')
data = allData.iloc[:, 0]
original_new = data[234:]
data = data[0:234]
order = st.arma_order_select_ic(data, ic=['aic', 'bic'])
model = ARMA(data, order=(4, 3))
result_arma = model.fit(disp=-1, method='css')
print(result_arma.summary())
predict_ts = result_arma.predict()
err = (data - predict_ts).dropna()
p_value = acorr_ljungbox(err, [6, 12, 18, 24])
print(p_value)
predict_new = result_arma.predict(
234,
271,
)
ax = predict_new.plot(label='forecast')
original_new.plot(label='observed')
ax.set_xlabel('Month')
ax.set_ylabel('Precipitation')
plt.legend()
plt.show()
predict_new = predict_new.values
original_new = original_new.values
tp = 0
tn = 0
