###############

#

# Transform R to Python Copyright (c) 2019 Masahiro Imai Released under the MIT license

#

###############

import os

import matplotlib.pyplot as plt

import pystan

import pandas

import pickle

import seaborn as sns

import tools

sales_df = pandas.read_csv('5-3-1-sales-ts-1-NA.csv')

print(sales_df.head())

print(sales_df.describe())

ax = sns.lineplot(x="date", y="sales", data=sales_df)

plt.show()

# drop NaN

sales_df_dropna = sales_df.dropna(how='any')

print(sales_df_dropna.head())

print(sales_df_dropna.describe())

print(len(sales_df['sales']))

print(len(sales_df_dropna['sales']))

# is Null

print(sales_df.isnull())

# getting not null index

print(sales_df.loc[sales_df.notnull()['sales']].index.values)

sales = sales_df['sales']

T = len(sales)

y = sales_df.dropna(how='any')['sales'].values

obs_no = sales_df.loc[sales_df.notnull()['sales']].index.values + 1

len_obs = len(obs_no)

print(len_obs)

print(obs_no)

print(y)

stan_data = {

}

if os.path.exists('5-3-2-local-level-interpolation.pkl'):

# sm = pickle.load(open('5-3-2-local-level-interpolation.pkl', 'rb'))

sm = pystan.StanModel(file='5-3-2-local-level-interpolation.stan')

else:

# a model using prior for mu and sigma.

sm = pystan.StanModel(file='5-3-2-local-level-interpolation.stan')

control = {

'adapt_delta': 0.8,

'max_treedepth': 10

}

mcmc_result = sm.sampling(

)

print(mcmc_result)

mcmc_result.plot()

plt.show()

# saving compiled model

if not os.path.exists('5-3-2-local-level-interpolation.pkl'):

with open('5-3-2-local-level-interpolation.pkl', 'wb') as f:

pickle.dump(sm, f)

mcmc_sample = mcmc_result.extract()

# plot ssm

tools.plot_ssm(mcmc_sample,

sales_df['date'],

'local level model',

'sales',

'mu',

sales_df['sales'])

# plot ssm about prediction

tools.plot_ssm(mcmc_sample,

sales_df['date'],

'local level model',

'sales',

'y_pred',

sales_df['sales'])