The “AI, ML and Big Data in Travel and Hospitality” webinar was hosted by:
- Andrew Sanders, VP, Travel, Transportation & Hospitality, DataArt
- Stan Boyer, Expert in Airline, Travel and IT, Advisor, DataArt
- Vlad Bessmertnyy, AI Consultant, Leader of Financial AI, DataArt
- Dmitry Baykov, Senior ML Engineer/Data Scientist, DataArt
Our experts revealed a flexible AI/ML solution for increasing incremental revenue and improving traveler experience.
The Challenge
Setting up a project requires data, which produces even more data in return. To get value from your projects, especially in the case of machine learning (ML), companies need to handle substantial volumes of data, prioritizing which information to use. The main pain point is that the standard data funnel requires significant human input and manual data processing. That is why developing an automated tool to optimize the entire process is a challenge worth facing.
The Solution
DataArt's solution, which was developed by experts Vlad Bessmertnyy and Dmitry Baykov, began with participation in the Kaggle Competition. Initially, DataArt's experts had to proceed with multiple stages of a common machine learning flow. On average, it takes 10-15 days to complete the stages. Our team's core idea was to create a more efficient and fully-automatic ML solution that requires only one manual step to set up a fully automated data pipeline. This ambitious goal inspired Vlad Bessmertnyy and Dmitry Baykov to develop a tool to optimize human input and expand variations.
After implementing and applying the AutoML solution, DataArt's team achieved impressive accuracy with its result, reducing the time required to obtain benchmark results to 3-5 days (a 3x improvement over traditional methods). Performance was improved after DataArt added an additional (but optional) collaborative step with its data scientists, which occurred in addition to receiving AutoML usage results. Optimizing this model gave even higher scores on the Kaggle competition leaderboard.
AutoML Advantages:
- Advanced starting point for a data science team
- Significant time savings for data scientists (up to 3x reduction in time spent)
- Feature engineering focused on multi-relational data tables
- Web app and SDK versions available
- Scalable approach to feature generation (Kubernetes)
- Cloud-agnostic solution can run on-premises or in the cloud (GCP, AWS, or Azure)
- Domain-agnostic approach to machine learning in travel and hospitality, finance, insurance, media, and healthcare
- Compatible with the most popular ML model types.
Advanced feature engineering with an AutoML Pipeline lets you combine several data sources with multiple relations into a single dataset with new features. The tables and relations between these tables prepared with the tool can provide useful datasets for machine learning tasks like Recommendations, Similarity Analysis, Anomaly Detection, Churn Prediction, Risk Analysis, and Sales and Fraud Prediction.
What is in it for Travel?
After a user decides to plan a journey and makes all preparations after the actual purchase, there is a period between the initial purchase and starting the travel, which can potentially benefit both the travel business and the user. We call this time gap the "Idle Space." Currently, companies seem to be unsure how to interact with users during this period.
The "Idle Space" can be the perfect time for AutoML to get into the game and to begin targeting relevant offers that are specially tailored for that customer.
What Is the Benefit to Hospitality Companies?
Hospitality is a traditionally people-oriented sector that must reconcile a considerable paradigm shift toward automation without losing some of the basics usually associated with a great personal experience. In the recent past, the term "luxury hotel" was almost defined as a place where guests could have great facilities and technology in their rooms. Now offering better service means having technology behind the scenes, proactively anticipating guests’ needs.
However, there is some disagreement on this point. Most recent surveys find that customers prefer self-service options for easy answers. However, they still want interaction with a human if they cannot handle the issue with a snap of a finger.
Custom Software for Your Travel Business
Learn MoreHow do Industry Leaders Harness this Science?
Marriott Hotels partnered with Alibaba to simplify the check-in process using facial recognition, which is being used today in China. Volara by Cloud5 and Angie by Nomadix are two common tools to deliver AI-powered voice interaction for guests in the room. Ivy by Go Moment is one of the leading AI-based concierge systems. With its AI-powered chatbot, Radisson Blu uses AI to give directions, handle complaints, and order room service. However, the No. 1 machine-learning initiative in hospitality are recommendations search engines like Expedia.
Machine Learning and Data Science
Learn MoreConclusion
The main notable trends in the industry are a convergence of voice and screen, mobile-based interaction, increased use of resourcing systems that optimize staffing and energy, improved personalized, targeted offers, and use of machine learning to stop cybercrime. If you decide that you use AI/ML for business, you do not need a large amount of data to get started. Define business objectives and start conducting experiments with the data you already have. Finally, challenge your internal team and vendors to develop innovative approaches to using big data — it will pay off in the long term.
The DataArt team has helped numerous companies in the travel and hospitality industry develop and apply various custom technological solutions. If you are looking for technological expertise and travel domain knowledge, contact us.
