Download and plot seatfinder data

seatfinder-download-and-plot.py

import requests
import os

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

BASE_URL = 'https://seatfinder.bibliothek.kit.edu/kassel/getdata.php?location%5B0%5D=UBA0EP,UBA1EP,UBA2EP,UBA3EP,UBB0EP,UBB0GP,UBB1EP,UBB1GP,UBB2EP,UBB2GP,LeoEG,LeoOG&values%5B0%5D=manualcount,seatestimate&after%5B0%5D=-1year&before%5B0%5D=now'
DELIMITER = ';'
FILE_MANUAL_COUNT = 'seatfinder_manual_count.csv'
FILE_SEAT_ESTIMATE = 'seatfinder_seat_estimate.csv'
PLOT_X_INCHES = 20
PLOT_Y_INCHES = 40


def parse_seatfinder_json(snippet):
    """Parses a hierarchical json snippet from seatfinder and returns a list of records."""
    ret = [['location_name', 'timestamp', 'occupied_seats', 'free_seats']]
    for k1, v1 in snippet.items():
        for e in v1:
            ret.append([
                e['location_name'],
                e['timestamp']['date'],
                str(e['occupied_seats']),
                str(e['free_seats']),
            ])
    return ret


def save_csv(snippet, file_name):
    data = parse_seatfinder_json(snippet)
    csv = '\n'.join([';'.join(e) for e in data])
    with open(file_name, 'w') as text_file:
        text_file.write(csv)


def download_seatfinder_data():
    r = requests.get(BASE_URL)
    data = r.json()[0]
    save_csv(data['manualcount'], FILE_MANUAL_COUNT)
    save_csv(data['seatestimate'], FILE_SEAT_ESTIMATE)


# Download data if not found
for file_name in [FILE_MANUAL_COUNT, FILE_SEAT_ESTIMATE]:
    if not os.path.isfile(file_name):
        download_seatfinder_data()
        break

# Load data
df_manual_count = pd.read_csv(FILE_MANUAL_COUNT, delimiter=DELIMITER, parse_dates=['timestamp'])
df_seat_estimate = pd.read_csv(FILE_SEAT_ESTIMATE, delimiter=DELIMITER, parse_dates=['timestamp'])
locations = df_manual_count.location_name.unique()

# Create subplots with shared x axis
fig, axes = plt.subplots(len(locations), sharex=True, sharey=True)
fig.set_size_inches(PLOT_X_INCHES, PLOT_Y_INCHES)
fig.suptitle('Estimated counts')

# Save this for later limits
dates = pd.to_datetime(df_seat_estimate.timestamp, format='%Y-%m-%d %H:%M:%S.%f')
colors = ["#1f77b4", "#ff7f0e", "#2ca02c", "#d62728", "#9467bd", "#8c564b", "#e377c2", "#7f7f7f", "#bcbd22", "#17becf"]

for i, location in enumerate(locations):
    ax = axes[i]
    color1 = colors[i % len(colors)]
    color2 = colors[(i + 1) % len(colors)]
    df_manual_count_filtered = df_manual_count[df_manual_count['location_name'] == location]

    # Plot estimated count as connected lines
    df_seat_estimate_filtered = df_seat_estimate[df_seat_estimate['location_name'] == location]
    g1 = sns.lineplot(x='timestamp', y='occupied_seats', color=color1, data=df_seat_estimate_filtered, ax=ax)
    g1.set_xlabel('Date/Time')
    g1.set_ylabel('Occupied seats')

    # Plot manual count as dots
    g2 = sns.scatterplot(x='timestamp', y='occupied_seats', color=color2, data=df_manual_count_filtered, ax=ax)
    g2.set_xlabel('Date/Time')
    g2.set_ylabel('Occupied seats')

    ax.set_xlim(dates.min(), dates.max())

# Uncomment to view interactively
# plt.show()
plt.savefig('seatfinder')