forked from h2oai/driverlessai-recipes
-
Notifications
You must be signed in to change notification settings - Fork 0
/
days_until_dec2020.py
44 lines (34 loc) · 1.53 KB
/
days_until_dec2020.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
"""Creates new feature for any date columns, by computing the difference in days between the date value and 31st Dec 2020"""
# days_until_dec2020.py
# Transformer creates an "Days Until" feature based on date columns. "Days Until"
# being the no. of days before 2020-12-31. You can change that date by modifying code below
#
# We can also substitute the fixed data with "today's date", but the function
# will become quite mutable and will provide unpredictable results each day.
# Just pick a target date, change the name of the function after 2020 and use it as a transformer
#
from h2oaicore.transformer_utils import CustomTransformer
import datatable as dt
import numpy as np
import pandas as pd
import dateparser
_global_modules_needed_by_name = ['regex==2018.1.10', 'dateparser==0.7.1']
def convert_to_age(ts):
if (type(ts) == "date"):
time1 = dateparser.parse(ts)
time2 = dateparser.parse("2020-12-31 11:59:59")
# print(str(time1), str(time2), str((time2 - time1).days))
return (time2 - time1).days
else:
return (-1)
class DaysUntilDec2020(CustomTransformer):
@staticmethod
def get_default_properties():
return dict(col_type="date", min_cols=1, max_cols=1, relative_importance=1)
def fit_transform(self, X: dt.Frame, y: np.array = None):
return self.transform(X)
def transform(self, X: dt.Frame):
if (X.nrows == 0):
return X.to_pandas().iloc[:, 0]
else:
return X.to_pandas().apply(lambda row: convert_to_age(row[0]), axis=1)