Philipp Bohnenstengel phibo23

Analyzing Nextbike Data

The dataset consists of a huge directory of json files

each file contains a snapshot of https://api.nextbike.net/maps/nextbike-live.json?countries=de
filenames are ISO8601 timestamps, except the : are replaced with _

My goal was to create timeseries for the amount of available and booked bikes in Karlsruhe.

DuckDB (https://duckdb.org/) was used to convert data from the list of JSON files to a single csv file. Running the following query on ~60GB of files on a USB flash drive took about 5 minutes on a 2021 MacBook pro (M1 Pro CPU, 32 GB RAM)

	<?php
	/*
	original code by cogitom: https://gist.github.com/997980/

	This script reads future events (plus several days into the past) from Facebook pages
	and creates a subscribable iCalendar

	Improvements upon the original version:
	- fixed start and endtime reading
	- timezone adjustment done on facebook's side (more reliable)

	// You can use the entire C/C++ standard library, just add the relevant
	// #includes. We recommend math.h ;-)
	#include <algorithm>
	#include <string>
	#include "usercode.h"

	/*
	* This is your bot's startup function. Here you can set your snake's colors,
	* set up persistent variables, etc.
	*/

	const widget = new ListWidget()

	const request = new Request('https://iceportal.de/api1/rs/status')
	// {"connection":true,"serviceLevel":"AVAILABLE_SERVICE","gpsStatus":"LAST_KNOWN_POSITION","internet":"MIDDLE","latitude":50.9862425,"longitude":9.572488166666666,"tileY":109,"tileX":-2,"series":"412","serverTime":1704210660163,"speed":189.0,"trainType":"ICE","tzn":"ICE9474","wagonClass":"FIRST","connectivity":{"currentState":"HIGH","nextState":"UNSTABLE","remainingTimeSeconds":4800},"bapInstalled":true}
	const json = await request.loadJSON()
	const { speed } = json
	widget.addText(speed?.toLocaleString('de', {
	style: 'unit',
	unit: 'kilometer-per-hour'
	}))