yearlystats.py

import calendar
import csv
import datetime
import dateutil.parser
import json
from dateutil.rrule import rrule, MONTHLY, YEARLY
from urllib import request
from urllib.parse import urlencode

base_url = "https://www.performance.service.gov.uk/data/register-to-vote/volumetrics"

start_date = datetime.datetime(2015,1,1)
end_date = datetime.datetime(2019,12,31)

age_groups = [
    "18-24",
    "25-34"
]
metrics = {}

f = open("./metrics.csv", "w")
writer = csv.writer(f)

headers = ["Year"]

for date in rrule(MONTHLY, dtstart=start_date, until=end_date):

    headers.append(date.month)

    (_, last) = calendar.monthrange(date.year, date.month)

    start_at = datetime.datetime(date.year, date.month, 1).astimezone(datetime.timezone.utc)
    end_at = datetime.datetime(date.year, date.month, last, 23, 59, 59).astimezone(datetime.timezone.utc)

    qs = urlencode({
        'period': 'month',
        'filter_by': 'metricName:age_band',
        'collect': 'count:sum',
        'group_by': 'value',
        'start_at': start_at.isoformat(),
        'end_at': end_at.isoformat(),
        'format': 'json',
    })

    url = f'{base_url}?{qs}'
    print(f'Making request to {url}')

    req =  request.Request(base_url + '?' + qs)
    resp = request.urlopen(req)

    payload = json.load(resp)
    data = payload["data"]
    key = f'{date.year}-{date.month}'

    metrics[date.year] = metrics.get(date.year) or {}

    for segment in data:

        if segment.get('value') in age_groups:

            year_stats = metrics[date.year]

            year_stats[date.month] = year_stats.get(date.month) or 0
            year_stats[date.month] += segment.get('count:sum')

writer.writerow(headers)

for year, value in metrics.items():

    values = list(value.values())

    row = [year] + values
    writer.writerow(row)