import pandas as pd
import numpy as np
from datetime import datetime, timedelta
from statsmodels.tsa.statespace.sarimax import SARIMAX
from scipy.stats import norm
dataset = dataset[['Start of Week','23756856']]
def datetimeix(dataset):
dataset['Start of Week'] = pd.DatetimeIndex(dataset['Start of Week'])
dataset.set_index('Start of Week', inplace = True)
return dataset
dataset = datetimeix(dataset)
data_train = dataset[:45]
data_for_dist_fitting = dataset[45:]
history = np.log(dataset.astype(float))
predictions= []
train_end =datetime(2022,2,28)
rolling_predictions = data_for_dist_fitting.copy()
i=0
for train_end in data_for_dist_fitting.index+timedelta(7):
train_data = history[:train_end+timedelta(7)]
model = SARIMAX(train_data,order=(1,1,1),seasonal_order=(1, 1, 1, 12),enforce_invertibility=False, enforce_stationarity=False)\
model_fit = model.fit()
output = model_fit.predict()
yhat = np.exp(output)
rolling_predictions[train_end] = yhat
predictions[i]= rolling_predictions.iloc[i, i+1]
i=i+1
predictions = pd.DataFrame({'Date': data_for_dist_fitting.index ,'#Prediction':predictions })
