def _get_feries(self) -> List[datetime]:
"""Renvoie la liste des jours fériés de l'année x et de l'année x +1
:return: liste des jours fériés
:rtype: List[datetime]
"""
days = []
for year in (self.year, self.year + 1):
days += [_[0] for _ in France().holidays(year)]
self.feries = days
return days
def is_day_off(date, weekday):
cal = France()
splitted = ((date.split(" "))[0]).split("-")
#print(splitted)
y = int(splitted[0])
m = int(splitted[1])
d = int(splitted[2])
if (cal.is_working_day(dt.date(y,m,d)) and (weekday != 6)):
return 0
else:
return 1
def fill_not_working_days(self):
""" Create columns WEEK_END and DAY_OFF from DATE """
# days-off
cal = France()
self.df["DAY_OFF"] = np.vectorize(cal.is_holiday)(
self.df.index.to_pydatetime()).astype(int)
# week-end
self.df["WEEK_END"] = np.in1d(self.df.index.dayofweek,
[5, 6]).astype(int)
# not working day
self.df["NOT_WORKING_DAY"] = self.df["DAY_OFF"] | self.df["WEEK_END"]
# daytime
self.df["DAYTIME"] = self.df.index.map(self.__is_daytime)
def est_ouvr(_dt) : # Imports from app.models import TDateFermeture from workalendar.europe import France output = False # Initialisation d'un calendrier français cal = France() # Vérification de l'ouvrabilité if cal.is_working_day(_dt) == True and TDateFermeture.objects.filter(pk = _dt).count() == 0 : output = True return output
def __init__(self, _date):
self.date = _date
# if required, _calendar.is_working_day can deal with extra working days/holidays
self.is_working_day = France().is_working_day(_date)
self.weekday = _date.weekday()
# normal working day if no holiday or if weekend day
# tip: holiday during weekend are considered as normal weekend working day
if self.weekday in [5, 6] or self.is_working_day:
self.profile = copy.deepcopy(Week.SLOTS[self.weekday])
# unusual working day if holiday
else:
self.profile = copy.deepcopy(Week.SLOTS[Week.KEY_SPECIAL_HOLIDAY])
if not self.profile:
raise UserWarning(
"daily planning's profile has not been successfully resolved.")
def get_cal_from_country(bmk_country):
"""
Function returning a calendar based on the 'benchmark_country' of the csv file
# Python package to manage holidays per country
# >> See : https://github.com/peopledoc/workalendar
from 'benchmark_country' column (to be parsed) in the Derivation Script
Warning : Tuples may appear like [USA, Japon] or [USA, China] instead of China
[Germany, China] instead of Russia
[USA, China] instead of Moyen-Orient
[USA, China] instead of Brasil
Currently missing : China, Russia
@TODO : ADD HONG-KONG !!! (for 'HSI_Index')
NOTE : 5 avril 2018 : Ching Ming Festival (jour férié Hong-Kong !)
:param bmk_country: benchmark country (type: string)
:return:
- cal: calendar related to the country (type: workalendar type ?)
"""
cal = []
if ',' in bmk_country: # '[A, B]' => ['A', 'B']
print "[WARNING] Tuple for the 'benchmark_country : {}, returning the first one..".format(
bmk_country)
bmk_country = bmk_country.replace('[', '').replace(']', '').split(',')
bmk_country = bmk_country[0] # TO BE DEFINED !
if bmk_country == 'USA':
cal = UnitedStates()
elif bmk_country == 'Germany':
cal = Germany()
elif bmk_country == 'Japan':
cal = Japan()
elif bmk_country == 'France':
cal = France()
elif bmk_country == 'UK':
cal = UnitedKingdom()
elif bmk_country == 'Grèce':
cal = Greece()
elif bmk_country == 'Italie':
cal = Italy()
elif bmk_country == 'Espagne':
cal = Spain()
elif bmk_country == 'Brasil':
cal = Brazil()
return cal
def get_nb_half_days_by_period_per_month(date_from, date_to):
day_list = pd.date_range(date_from, date_to).tolist()
cal = France()
holidays = cal.holidays(date_from.year)
holidays_date = [h[0] for h in holidays]
md_dict = {}
for i in range(1, 13):
md_dict[i] = {
"monday_am": 0,
"monday_pm": 0,
"tuesday_am": 0,
"tuesday_pm": 0,
"wednesday_am": 0,
"wednesday_pm": 0,
"thursday_am": 0,
"thursday_pm": 0,
"friday_am": 0,
"friday_pm": 0,
}
# for each day of the period
for day in day_list:
if not day.date() in holidays_date:
if day.dayofweek == 0:
md_dict[day.month]['monday_am'] += 1
md_dict[day.month]['monday_pm'] += 1
if day.dayofweek == 1:
md_dict[day.month]['tuesday_am'] += 1
md_dict[day.month]['tuesday_pm'] += 1
if day.dayofweek == 2:
md_dict[day.month]['wednesday_am'] += 1
md_dict[day.month]['wednesday_pm'] += 1
if day.dayofweek == 3:
md_dict[day.month]['thursday_am'] += 1
md_dict[day.month]['thursday_pm'] += 1
if day.dayofweek == 4:
md_dict[day.month]['friday_am'] += 1
md_dict[day.month]['friday_pm'] += 1
return md_dict
def clean_electricity(self):
import datetime
#Get holiday dates in France using the library called workalendar
from workalendar.europe import France
cal = France()
#Convert dates to DateTime format
self.df['Date'] = pd.to_datetime(self.df['Date'], format='%d/%m/%Y')
#Create the DayTypes mentioned in the article (except DayType 8)
#/!\ We chose an arbitrary order to create the day types, since they are not separated sets
#Add normal weekdays (DayType 1)
self.df['Day_type'] = self.df['Date'].apply(lambda u: (u.date(), 1))
#Add Mondays (DayType 0)
self.df['Day_type'] = self.df['Day_type'].apply(
lambda x: (x[0], 0) if x[0].weekday() == 0 else x)
#Add Fridays (DayType 2)
self.df['Day_type'] = self.df['Day_type'].apply(
lambda x: (x[0], 2) if x[0].weekday() == 4 else x)
#Add Saturdays (DayType 3)
self.df['Day_type'] = self.df['Day_type'].apply(
lambda x: (x[0], 3) if x[0].weekday() == 5 else x)
#Add Sundays (DayType 4)
self.df['Day_type'] = self.df['Day_type'].apply(
lambda x: (x[0], 4) if x[0].weekday() == 6 else x)
#Add Bank holidays (DayType 6)
self.df['Day_type'] = self.df['Day_type'].apply(
lambda x: (x[0], 6) if cal.is_holiday(x[0]) == True else x)
#Add Before Bank holidays (DayType 5)
self.df['Day_type'] = self.df['Day_type'].apply(
lambda x: (x[0], 5) if cal.is_holiday(x[0] + datetime.timedelta(
days=1)) == True else x)
#Add After Bank holidays (DayType 7)
self.df['Day_type'] = self.df['Day_type'].apply(
lambda x: (x[0], 7) if cal.is_holiday(x[0] - datetime.timedelta(
days=1)) == True else x)
self.df['Day_type'] = self.df['Day_type'].apply(lambda x: x[1])
self.df['Date'] = self.df['Date'].apply(lambda u: u.date())
]
name_encoder = dict(zip(names, [list(x) for x in np.eye(len(names))[:, 1:]]))
weekdays = [
'Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday',
'Sunday'
]
# features
features = (
['YEAR', 'MONTH', 'DAY', 'WEEKEND', 'HOLIDAY', 'TIMESLOT', 'DAYSLOT'] +
weekdays[:-1] + names[1:] + ['CALLS'])
# French calendar object
calendar = France()
# first and last dates possibly considered
first = datetime.strptime('2011-01-01 00:00:00', '%Y-%m-%d %H:%M:%S')
last = datetime.strptime('2013-12-31 23:30:00', '%Y-%m-%d %H:%M:%S')
def parse_date(datestring):
""" Parse date string and return relevant features. """
# datetime object encoding data information
date = datetime.strptime(datestring[:-4], '%Y-%m-%d %H:%M:%S')
# create date-related predictor
weekday = [0] * 6
import daiquiri
import numpy as np
import pandas as pd
from dateutil import parser
from workalendar.europe import France
from sklearn.cluster import KMeans
import xgboost as xgb
from jitenshea import config
daiquiri.setup(logging.INFO)
logger = daiquiri.getLogger("stats")
# French Calendar
cal = France()
SEED = 2018
np.random.seed(SEED)
CLUSTER_ACT_PATH_CSV = 'jitenshea/data/cluster_activite.csv'
###################################
### CLUSTER ACTIVITE
###################################
def preprocess_data_for_clustering(df):
"""Prepare data in order to apply a clustering algorithm
Parameters
#! /usr/bin/env python
# -*- coding: utf-8 -*-
"""Use weekalendar to generate holidays."""
from collections import OrderedDict
from workalendar.europe import France
holidays = []
for year in range(1990, 2020):
holidays += France().get_calendar_holidays(year)
header = """#! /usr/bin/env python
# -*- coding: utf-8 -*-
from datetime import datetime
holidays = ["""
footer = """
]
"""
with open("../assets/holidays.py", "w") as text_file:
text_file.write(header)
for holiday_date, holiday_name in OrderedDict(holidays).iteritems():
text_file.write("""
datetime.strptime("{}", "%Y-%m-%d").date(), # {}""".format(
holiday_date, holiday_name))
text_file.write(footer)
def is_holiday_fr(date_of_interest):
cal = France()
result = cal.is_holiday(date_of_interest) # this will return a boolean True/False value
# cast to integer
result = int(result)
return result
import pandas as pd
from workalendar.europe import France
CAL = France()
map_jour_fr = {
'JOHV': 'Jour Ouvré Hors Vacances Scolaires',
'SAHV': 'Samedi Hors Vacances Scolaires',
'JOVS': 'Jour Ouvré en période de Vacances Scolaires',
'SAVS': 'Samedi en période de Vacances Scolaires',
'DIJFP': 'Dimanche, Jour Férié et ponts' # TODO:pont
}
def is_dimanchelike(day):
"""Dimanche, Jour Férié
"""
return not CAL.is_working_day(day)
def read_vacances(path):
mapping = {
'Académies': 'academie',
'Date de début': 'begin',
'Date de fin': 'end',
'Description': 'description'
}
df = (pd.read_csv(path).rename(columns=mapping))
for d in ['begin', 'end']:
df[d] = pd.to_datetime(df[d])
def __init__(self, school_year):
"""Creates the class
:param school_year: Starting year for the school year
:type school_year: int
"""
# Initialize school calendar
school_calendar = {}
# Get holidays from National Education Ministry
with open("Calendrier_Scolaire_Zone_B.ics", "r") as cal:
edu_calendar = icsCalendar(cal.read())
# Initialize days off ICS calendar
days_off_school = icsCalendar()
# Define begin/end of the school year we are in
for event in edu_calendar.events:
if ("Rentrée scolaire des élèves" in event.name
and event.begin.year == school_year):
school_begin = event.begin
continue
if "Vacances d'été" in event.name and event.begin.year == school_year + 1:
school_end = event.begin
days_off_school.events.add(event)
continue
# Add events between begin and end of school year
# First add school holidays, starting the day after the beginning
for event in edu_calendar.events:
if event.begin > school_begin and event.begin < school_end:
# Remove 1s to end so that it does not overrides next day
event._end_time = event._end_time.shift(seconds=-1)
days_off_school.events.add(event)
# Then take holidays from workalendar
calendar = France()
for year in [school_year, school_year + 1]:
for day_off in calendar.holidays(year):
begin = day_off[0]
event = Event(day_off[1], begin)
event.make_all_day()
if event.begin > school_begin and event.begin < school_end:
# Remove 1s to end so that it does not overrides next day
event.end = event.end.shift(seconds=-1)
days_off_school.events.add(event)
# Create timeline object
timeline = days_off_school.timeline
def get_last_tue(date):
while date.weekday() != 1:
date = date.shift(days=-1)
return date
def set_end(date):
return date.replace(hour=16, minute=30, second=0, microsecond=0)
# Now create dict of all days of school year and status
# ordered by week number
current = school_begin
while int(current.strftime("%w")) > 1:
current = current.shift(days=-1)
while current < school_end:
day = current.strftime("%d")
week_number = current.isocalendar()[1]
if week_number not in school_calendar.keys():
school_calendar[week_number] = []
current_week = school_calendar[week_number]
# Limite cantine
limit = set_end(
get_last_tue(
current.shift(
weeks=-1) if current.weekday() >= 1 else current))
past_cantine = arrow.now() > limit
past_garderie = current < arrow.now()
month = current.strftime("%B").capitalize()
data = {
"day": day,
"month": month,
"date": current.strftime("%Y%m%d")
}
# If one of these is past, then save immediately
data["bookable_cantine"] = not past_cantine
data["bookable_garderie"] = not past_garderie
if not data["bookable_cantine"] and not data["bookable_garderie"]:
current_week.append(data)
current = current.shift(days=+1)
continue
if current.isoweekday() in NON_WORKING_DAYS or current.isoweekday(
) in [
6,
7,
]:
data["bookable_cantine"] = False
data["bookable_garderie"] = False
data["bookable"] = False
current_week.append(data)
elif len([i for i in timeline.at(current)]) > 0:
data["bookable_cantine"] = False
data["bookable_garderie"] = False
data["bookable"] = False
current_week.append(data)
elif current < school_begin:
data["bookable_cantine"] = False
data["bookable_garderie"] = False
data["bookable"] = False
current_week.append(data)
else:
data["bookable"] = True
current_week.append(data)
current = current.shift(days=+1)
# Create set of periods
i = 1
periods = {
i: {
"begin": school_begin.format("YYYYMMDD"),
"begin_pretty": school_begin.format("DD/MM/YYYY"),
}
}
for event in sorted(edu_calendar.events):
if event.begin < school_begin:
continue
if event.begin > school_end:
continue
if "Vacances" in event.name:
periods[i]["end"] = event.begin.format("YYYYMMDD")
periods[i]["end_pretty"] = event.begin.format("DD/MM/YYYY")
i += 1
periods[i] = {
"begin": event.end.format("YYYYMMDD"),
"begin_pretty": event.end.format("DD/MM/YYYY"),
}
periods.pop(i)
self.periods = periods
self.calendar = school_calendar
# from fbprophet import make_holidays as mhdays import json from capacity_planning.forecast.utilities.holidays import events_and_hols as evh from capacity_planning.utilities import sys_utils as s_ut FILE_PATH = os.path.dirname(os.path.abspath(__file__)) country_hols = dict() from workalendar.usa import UnitedStates country_hols['UnitedStates'] = UnitedStates() from workalendar.europe import Russia country_hols['Russia'] = Russia() from workalendar.europe import France country_hols['France'] = France() from workalendar.europe import Belgium country_hols['Belgium'] = Belgium() from workalendar.europe import Spain country_hols['Spain'] = Spain() from workalendar.europe import Germany country_hols['Germany'] = Germany() from workalendar.europe import Austria country_hols['Austria'] = Austria() from workalendar.europe import Italy country_hols['Italy'] = Italy() from workalendar.europe import Portugal country_hols['Portugal'] = Portugal() from workalendar.europe import UnitedKingdom country_hols['UnitedKingdom'] = UnitedKingdom() from workalendar.europe import Ireland