#Estimation Time Series Decomposition (ETS)
import numpy as np
import matplotlib.pyplot as plt
from sklearn.neighbors import KNeighborsClassifier
from statsmodels.tsa.seasonal import seasonal_decompose
import pandas as pd
from pmdarima import auto_arima
dataset=pd.read_csv("AirPassengers.csv",index_col='Month',parse_dates=True)
result=seasonal_decompose(dataset['#Passengers'],model='multiplicative')
result.plot()
plt.show()
print(dataset)
result=auto_arima(dataset['#Passengers'],start_p=1,start_q=1,max_p=3,max_q=3,m=12,start_P=0,seasonal=True,d=None,D=1,trace=True,stepwise=True)
#SARIMAX from statsmodels.tsa.statespace.sarimax import SARIMAX train=dataset.iloc[:len(dataset)-12] test=dataset.iloc[len(dataset)-12:] model=SARIMAX(train['#Passengers'],order=(0,1,1),seasonal_order=(2,1,1,12)) result=model.fit() result.summary() start=len(train) end=len(train)+len(test)-1 pred=result.predict(start,end,typ='levels').rename('prediction') pred.plot(legend=True) test['#Passengers'].plot(legend=True) plt.show()