def ARMA_model(data, ohlc='Close'):
data = data[ohlc]
# choose best p, q parameters for our model using AIC optimization
params = bestParams(data)
model = ARIMA(data, order=(params[0], 0, params[2]))
res = model.fit()
#model_summary = res.summary().as_text()
model_summary = res.summary()
# write summary to file
#fileobj = open("quotes/static/model_results/ARMA_Summary.txt", 'w')
#fileobj.write(model_summary.as_text())
#fileobj.close()
fig, ax = plt.subplots(figsize=(10, 8))
ax = data.plot(ax=ax)
fig = plot_predict(res,
start=data.index[0],
end=data.index[-1],
ax=ax,
plot_insample=False)
legend = ax.legend(["Actual price", "Forecast", "95% Confidence Interval"],
loc='upper left')
fig.savefig("quotes/static/plots/forecast_vs_actual.jpg")
return (model, res, model_summary)
def test_predict_plot(use_pandas, model_and_args, alpha):
model, kwargs = model_and_args
rs = np.random.RandomState(0)
y = rs.standard_normal(1000)
for i in range(2, 1000):
y[i] += 1.8 * y[i - 1] - 0.9 * y[i - 2]
y = y[100:]
if use_pandas:
index = pd.date_range("1960-1-1", freq="M", periods=y.shape[0] + 24)
start = index[index.shape[0] // 2]
end = index[-1]
y = pd.Series(y, index=index[:-24])
else:
start = y.shape[0] // 2
end = y.shape[0] + 24
res = model(y, **kwargs).fit()
fig = plot_predict(res, start, end, alpha=alpha)
assert isinstance(fig, plt.Figure)
def Arima_And_plot(y, country):
arma_mod = ARIMA(y, order=(5, 0, 5), trend='n')
arma_res = arma_mod.fit()
SARIMAX_results = arma_res.summary()
start_date, end_date = max(y.index) - pd.Timedelta(days=90), max(
y.index) + pd.Timedelta(days=10)
st.write('## ARIMA model for {}'.format(country))
st.text(SARIMAX_results)
st.write('## Forecasts for {}'.format(country))
sns.set_style("darkgrid", {"axes.facecolor": ".9"})
f, ax = plt.subplots(figsize=(10, 8))
date_start = max(y.index) - pd.Timedelta(days=90)
ax.plot(y[(y.index >= date_start)], 'r', label='Actuals')
f = plot_predict(arma_res, start=start_date, end=end_date, ax=ax)
ax.legend(loc='upper right')
st.pyplot(f)
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from pandas import datetime
from statsmodels.graphics.tsaplots import plot_predict
from statsmodels.tsa.arima_process import arma_generate_sample
from statsmodels.tsa.arima.model import ARIMA
from sklearn.metrics import mean_squared_error
from math import sqrt
np.random.seed(0)
arparams = np.array([0.75, -0.25])
maparams = np.array([0.65, 0.35])
arparams = np.r_[1, -arparams]
maparams = np.r_[1, maparams]
nobs = 1000
y = arma_generate_sample(arparams, maparams, nobs)
dates = pd.date_range("1950-1-1", freq="M", periods=nobs)
y = pd.Series(y, index=dates)
arma_mod = ARIMA(y, order=(2, 0, 2), trend="n")
arma_res = arma_mod.fit()
print(arma_res.summary())
fig, ax = plt.subplots(figsize=(10, 8))
fig = plot_predict(arma_res, start="1999-06-30", end="2001-05-31", ax=ax)
legend = ax.legend(loc="upper left")
plt.show()
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from pandas import datetime
from statsmodels.graphics.tsaplots import plot_predict
from statsmodels.tsa.arima_process import arma_generate_sample
from statsmodels.tsa.arima.model import ARIMA
from sklearn.metrics import mean_squared_error
from math import sqrt
n = 100
N = 10000
mu = np.repeat(np.sin(np.arange(n)), int(N / n))
X = np.random.normal(5, 25, int(N))
Y = X + mu
plt.plot(Y)
plt.plot(mu, color='red')
plt.title("Ciclic trend")
plt.legend(['$Y_t$', '$\mu_t$'])
plt.show()
ma2 = ARIMA(Y, order=(0, 0, 2), trend="n")
ma2_res = ma2.fit()
print(ar1_res.summary())
fig, ax = plt.subplots(figsize=(10, 8))
fig = plot_predict(ma2_res, start=1, end=2 * N, ax=ax)
plt.plot(Y)
legend = ax.legend(loc="upper left")
plt.show()
e_t=np.random.normal(0,1) y=np.append(y,a1*y[-1]+a2*y[-2]+e_t) y.mean() y.std() #plt.plot(y);plt.show() # Predict l=20 ar2 = ARIMA(y, order=(2, 0, 0)) ar2_res = ar2.fit() print(ar2_res.summary()) fig, ax = plt.subplots(figsize=(10, 8)) fig = plot_predict(ar2_res, start=0,end=N+l, ax=ax) plt.plot(y) legend = ax.legend(loc="upper left") plt.show() #################################################################### # AR(1) N=1000 a1=0.5 sigma=0.1 x=np.arange(1) y=np.append(x,a1*x+np.random.normal(0,sigma))
