Introduction

Every tourist must pay a visitor's tax (Kurtaxe) for every night they are staying. The guests pay the tax directly to their host, who is responsible for correctly reporting the stays and relaying the collected taxes to the municipality.

Hotels must also report their overnight stays to the federal statistical office, which enables the municipalities to verify their reported stays rather easily. Leased apartments and houses have no further reporting responsibilities. The municipalities have therefore little to no tools to asses if the reported stays for these objects are correct.

The goal of this challenge is to develop data driven tools to facilitate the control of reported overnight stays.

Requirements

A comparison of visitor's tax with the previous year is to be made. For this, as many data sources as possible should be used to evaluate the overnight stays of the community Flumserberg. These should be compared with the paid visitor's tax.

Provided data

A list with all apartments inclusive number and size of rooms of Flums, one measurement per day of the wastewater data of the last 3 years in Flumserberg and a small part of the neighbouring area, 1035 measurements in total with precipitation and weather code and monthly electricity data for the municipality of Flumserberg from January 2017 to September 2019. Population figures for Flumserberg municipality at monthly level from February 2017 onwards

Deliverables

A dashboard with the key figures how big the deviation is, possibly which holiday homes are affected. A list of the holiday apartments that are obliged to pay visitor's tax and the number of overnight stays reported vs. the real number of overnight stays.

Update I: Chosen approaches

The project team explored and evaluated the existing data. The members decided on these methods:

1. Using a WebCrawler, the project team wants to check if all online posted houses/appartments are officially registered.

2. Collect and preprocess that data so it can be used for Machine learning.

3. Using a various linear models and random Forest, the project team wants to create a model to find inconstancies in the water consumption.

4. Using Time series analysis to identify tends, seasonality, in the behaviour of the heidiland region and identify any discrepencies.

Deliverables

Crawler that queries the Airbnb API to check if there are accomodatations that are posted online are registered at the municipitality

Script that prepocess the datasources provided by the tourism organization into a single data frame. Given that not all the data comes in the same granularity it is possible to

Time series analysis of electicity/water consumption versus over night stays

Data & Results

You can find all data, analyses and results under the button "Sources".