|
1
|
|
|
# --- |
|
2
|
|
|
# jupyter: |
|
3
|
|
|
# jupytext: |
|
4
|
|
|
# cell_metadata_json: true |
|
5
|
|
|
# formats: ipynb,py:percent |
|
6
|
|
|
# notebook_metadata_filter: language_info |
|
7
|
|
|
# text_representation: |
|
8
|
|
|
# extension: .py |
|
9
|
|
|
# format_name: percent |
|
10
|
|
|
# format_version: '1.3' |
|
11
|
|
|
# jupytext_version: 1.5.2 |
|
12
|
|
|
# kernelspec: |
|
13
|
|
|
# display_name: Python 3 |
|
14
|
|
|
# language: python |
|
15
|
|
|
# name: python3 |
|
16
|
|
|
# language_info: |
|
17
|
|
|
# codemirror_mode: |
|
18
|
|
|
# name: ipython |
|
19
|
|
|
# version: 3 |
|
20
|
|
|
# file_extension: .py |
|
21
|
|
|
# mimetype: text/x-python |
|
22
|
|
|
# name: python |
|
23
|
|
|
# nbconvert_exporter: python |
|
24
|
|
|
# pygments_lexer: ipython3 |
|
25
|
|
|
# version: 3.6.6 |
|
26
|
|
|
# --- |
|
27
|
|
|
|
|
28
|
|
|
# %% {"tags": ["parameters"]} |
|
29
|
|
|
# Our default parameters |
|
30
|
|
|
# This cell has a "parameters" tag, means that it defines the parameters for use in the notebook |
|
31
|
|
|
run_date = '2018-11-18' |
|
32
|
|
|
source_id = 'sensor1' |
|
33
|
|
|
nb_days = 7 |
|
34
|
|
|
|
|
35
|
|
|
# %% |
|
36
|
|
|
from datetime import datetime, timedelta |
|
37
|
|
|
import matplotlib.pyplot as plt |
|
38
|
|
|
import os |
|
39
|
|
|
import pandas as pd |
|
40
|
|
|
from pylab import rcParams |
|
41
|
|
|
import scrapbook as sb |
|
42
|
|
|
import statsmodels.api as sm |
|
43
|
|
|
|
|
44
|
|
|
|
|
45
|
|
|
# %% |
|
46
|
|
|
data_dir = "../data/input/step1" |
|
47
|
|
|
data = None |
|
48
|
|
|
run_datetime = datetime.strptime(run_date, '%Y-%m-%d') |
|
49
|
|
|
for i in range(nb_days): |
|
50
|
|
|
deltatime = run_datetime - timedelta(i) |
|
51
|
|
|
month_partition = deltatime.strftime("%Y-%m") |
|
52
|
|
|
delta = datetime.strftime(deltatime, '%Y-%m-%d') |
|
53
|
|
|
file = os.path.join(data_dir, month_partition, delta + "-" + source_id + ".csv") |
|
54
|
|
|
if os.path.exists(file): |
|
55
|
|
|
print("Loading " + file) |
|
56
|
|
|
new = pd.read_csv(file) |
|
57
|
|
|
if data is not None: |
|
58
|
|
|
data = pd.concat([data, new]) |
|
59
|
|
|
else: |
|
60
|
|
|
data = new |
|
61
|
|
|
|
|
62
|
|
|
# %% |
|
63
|
|
|
data['date'] = data['date'].apply(lambda x : datetime.strptime(x, "%Y-%m-%d %H:%M:%S")) |
|
64
|
|
|
print(data['date'].describe()) |
|
65
|
|
|
data.describe() |
|
66
|
|
|
|
|
67
|
|
|
# %% |
|
68
|
|
|
data = data.sort_values('date').set_index('date', drop=True) |
|
69
|
|
|
data = data.asfreq(freq="5min") |
|
70
|
|
|
data.head(5) |
|
71
|
|
|
|
|
72
|
|
|
# %% |
|
73
|
|
|
pred = sm.load("../data/input/step2/prediction_model_" + run_date + "-" + source_id) |
|
74
|
|
|
|
|
75
|
|
|
# %% |
|
76
|
|
|
pred_ci = pred.conf_int() |
|
77
|
|
|
rcParams['figure.figsize'] = 18, 8 |
|
78
|
|
|
fig, ax = plt.subplots() |
|
79
|
|
|
ax.plot(data[data.index > (run_datetime - timedelta(3))]['mydata'], label='observed') |
|
80
|
|
|
ax.plot(pred.predicted_mean, label='One-step ahead Forecast', alpha=.7) |
|
81
|
|
|
ax.fill_between(pred_ci.index, |
|
82
|
|
|
pred_ci.iloc[:, 0], |
|
83
|
|
|
pred_ci.iloc[:, 1], color='k', alpha=.2) |
|
84
|
|
|
ax.set_xlabel('Date') |
|
85
|
|
|
ax.set_ylabel('mydata') |
|
86
|
|
|
ax.set(title='Results on {}'.format(run_date)) |
|
87
|
|
|
fig.legend() |
|
88
|
|
|
|
|
89
|
|
|
# %% |
|
90
|
|
|
|
|
91
|
|
|
|
|
92
|
|
|
|
|
93
|
|
|
|
Aiscalate /
aiscalator
