Predictive pickup model for hotel revenue strategy
Summary of the entity:
The Data Cycle Hub (DCH) is the one-stop reference shop DIH in the Valencia region to foster data-driven and artificial intelligence-based innovation. The objective is to bridge the gap between research and industry, specifically SMEs, providing innovative solutions and services that require advanced data analytics, automatic learning, and artificial intelligence.
The DCH is a member of several DIH networks, such as EUHubs4Data (coordinator), AI DIH Network, DIH4CPS, INNDIH (E-DIH of the Valencia region, coordinator), and was one of the 5 finalists to the DIHNET DIH Champions Challenge 2019 and 2020. The DCH is coordinated by ITI – Instituto Tecnológico de Informática, a Research & Technology center specializing in ICT, located in Valencia, Spain. ITI provides a Big Data Space offering infrastructure, tools, and data for research and experimentation with Big Data Technologies, which has been awarded the BDVA i-Space Silver label and is a crucial infrastructure for the Data Cycle Hub.
This challenge has been defined together INVAT·TUR (Institute of Tourism Technologies of Region of the Valencia), a center conceived as a meeting platform for all agents in the tourism sector and represented one of the main axes in improving the competitiveness and sustainability of the tourism model of the Region of Valencia. The goals are to develop lines of action in R+D+i adapted to the needs of the tourism sector, as well as transfer knowledge to tourism companies and organizations, giving the tourism sector access to the most advanced knowledge, services, and technologies.
Summary of the challenge:
The challenge is to develop a predictive model for hotel properties that estimates future pickups (bookings that will be received in a given future period) including the observed and historical cancellation ratio, according to different traveler profiles.
Stakeholders: Hotel owners and travel operators
The pickup report is one of the data-based analysis tools used by hotel managers (and travel operators) to see the status of bookings received for a certain period and see: (a) the trend of bookings according to the analysis period (i.e. “X” weeks/months before the pickup period); and (b) the performance against historical data (e.g. rate of bookings in the previous year). This information is very useful because it enables the hotel owner to evaluate several aspects: (i) How far in advance did the bookings for a given period arrived last year)?; (ii) What is the performance of the booking trend of a certain period compared to the previous year?; (3) Did the modification of rates affected the bookings for the next period?; (4) On which weeks of the year (or days of the week) do the most bookings arrive?.
Most Hotel Property Management Systems include pickup reports, and some of them also predictive pickup models to generate these reports. However, these models do not include the cancellation ratio, which is another crucial variable that measures the speed at which the bookings are being canceled.
The objective of this challenge is to develop a predictive model that could be integrated into current Hotel Property Management Systems to enable the improvement of booking estimations according to the pickup modification rate AND the cancellation rate, for different traveler profiles (business, family, weekend). This predictive model should calculate the following. Given a future date period, the model should provide the estimation of potential bookings for this date, based on the previous year’s bookings and on the rate of observed bookings in the X previous months. This model must include the cancellation rate as a variable which would adjust the estimated bookings for the next period. The estimation of the booking period should be made for the next 52 weeks, with 6 months beforehand, and would depend on the profile of traveler selected.
The resolution of this challenge must be made in two steps. In a first step, the goal is to develop a Minimum Viable Product (MVP) of a pickup predictive model for flight bookings coming to Valencia region. Different datasets will be facilitated to build this model: flight searches, flight bookings (from countries of origin to Valencia airports), flight occupancy (nº of passengers that have finally flew to Valencia), and accommodation occupancy data (although taking into account that not all the flights are converted to hotel stays). From these data, an estimated flight pickup rate and flight cancellation rate could be extracted and used to generate an improved pickup predictive model. This MVP will serve to evaluate the visitors flow to the Valencia region for a given period, and also to showcase the potential to end-users, which will be involved in a second stage.
The visitor flow of tourists could be also used to evaluate the potential saturation of the territory in a given period, and thus, optimise the planning of resources in advance. This Territory Saturation index could be calculated as ((Sustainability offer/population) / (Sustainability offer/visitor flow))*100, where the Sustainability Offer is the sum of cultural heritage, natural heritage, and other tourist resources available within the territory.
In the second stage, a hotel will be involved in providing real data (bringing overnight stay and cancellation data) to generate the improved pickup model and validate the potential business model. In this stage, the initial MVP should be adapted to the data of the hotel and generate a representative pickup predictive dashboard.
REACH Data Provider:
- Turisme Comunitat Valenciana – INVAT.TUR: Different datasets covering the region of Valencia will be provided, under a private agreement with the applicants.
- Datasets for the first MVP (flights pickup model) will cover: flight capacity (by origin and target airport), flight prices (by origin and destination), flight bookings, searches by destination in a period of time (day, i.e. business profile; weekend, i.e. family profile; week, family). Flight data will include only legacy airline companies, but it could be used to generate the model for low-cost airline companies.
- Datasets for the second MVP (hotel pickup model) will cover: overnight stays and cancellation data of a hotel.
- Visitor flow associated to the Valencian territory.
- Open Source datasets:
- AENA Flight traffic management statistics. Includes data such as flight passengers, flight capacity, etc. for both legacy and low-cost airlines https://www.aena.es/es/estadisticas/consultas-personalizadas.html
- INE – Spain National Statistics Institute. Includes data such as national hotel booking statistics (by hotel type and month), cruise passengers (by data and port), ADR and RevPAR rates. https://www.ine.es/dyngs/INEbase/en/categoria.htm?c=Estadistica_P&cid=1254735576863
- Valencia Region data portal: including tourism data and others http://www.dadesobertes.gva.es/es/dataset
Spain open data portal: including tourism data and others https://datos.gob.es/en/catalogo?theme_id=turismo
At least one of the following should be addressed:
- Predictive pickup model of flight bookings coming to the Valencia region, adjusted with estimated flight cancellation rates
- Proof of concept of a predictive pickup model with cancellation rates per traveler profile for a given hotel and dashboard