1.solver：採用哪一種收斂演算法（不建議更改

2.disp：True會列印中間過程，我們直接設定False即可

3.transparam：預設是True，建議選擇true

Whether or not to transform the parameters to ensure stationarity. Uses the transformation suggested in Jones (1980). If False, no checking for stationarity or invertibility is done.

4.trend：=‘c'表示加一個截距項，=’nc‘表示不加截距項

舉個例子

from pandas import read_csv
from pandas import datetime
from matplotlib import pyplot
from statsmodels.tsa.arima_model import ARIMA
from sklearn.metrics import mean_squared_error
from math import sqrt
# load dataset
def parser(x):
	return datetime.strptime('190'+x, '%Y-%m')
series = read_csv('shampoo-sales.csv', header=0, parse_dates=[0], index_col=0, squeeze=True, date_parser=parser)
# split into train and test sets
X = series.values
size = int(len(X) * 0.66)
train, test = X[0:size], X[size:len(X)]
history = [x for x in train]
predictions = list()
# walk-forward validation
for t in range(len(test)):
	# fit model
	model = ARIMA(history, order=(4,1,0))
	model_fit = model.fit(disp=False, trend='c')
	print(model_fit.params)
	# one step forecast
	yhat = model_fit.forecast()[0]
	# store forecast and ob
	predictions.append(yhat)
	history.append(test[t])
# evaluate forecasts
rmse = sqrt(mean_squared_error(test, predictions))
print('Test RMSE: %.3f' % rmse)
 

from pandas import read_csv
from pandas import datetime
from matplotlib import pyplot
from statsmodels.tsa.arima_model import ARIMA
from sklearn.metrics import mean_squared_error
from math import sqrt
# load dataset
def parser(x):
	return datetime.strptime('190'+x, '%Y-%m')
series = read_csv('shampoo-sales.csv', header=0, parse_dates=[0], index_col=0, squeeze=True, date_parser=parser)
# split into train and test sets
X = series.values
size = int(len(X) * 0.66)
train, test = X[0:size], X[size:len(X)]
history = [x for x in train]
predictions = list()
# walk-forward validation
for t in range(len(test)):
	# fit model
	model = ARIMA(history, order=(4,1,0))
	model_fit = model.fit(disp=False, trend='nc')
	print(model_fit.params)
	# one step forecast
	yhat = model_fit.forecast()[0]
	# store forecast and ob
	predictions.append(yhat)
	history.append(test[t])
# evaluate forecasts
rmse = sqrt(mean_squared_error(test, predictions))
print('Test RMSE: %.3f' % rmse)
 

加截距項的結果：

...
[ 11.42283717  -1.16087885  -0.6519841   -0.547411    -0.28820764]
[ 11.75656838  -1.11443479  -0.61607471  -0.49084722  -0.24452864]
[ 11.40486702  -1.11705478  -0.65344924  -0.50213939  -0.25677931]
Test RMSE: 81.545

不加截距項的結果：

...
[-0.90717131 -0.22332019 -0.11240858 -0.04008561]
[-0.88836083 -0.21098412 -0.09046333 -0.02121404]
[-0.89260136 -0.24120301 -0.10243393 -0.03165432]
Test RMSE: 95.061

選哪一個有你自己決定

https://machinelearningmastery.com/tune-arima-parameters-python/

這一個文章有時序經典各種模型的cheatsheethttps://machinelearningmastery.com/time-series-forecasting-methods-in-python-cheat-sheet/