To get your personal travel recommendation, select several places you visited and can recommend to others.
Honolulu La Jolla Acapulco San Antonio Avalon Laguna Beach San Diego Scottsdale Havana Key West Miami Beach Miami Sarasota Pensacola Kissimmee Charleston Nice Portland West Hollywood Reno Seattle Kelowna Anaheim Los Angeles Atlanta Flagstaff Calgary Virginia Beach Atlantic City New York City Buenos Aires Lisbon Barcelona Bar Harbor Rome Venice Johannesburg Bangkok Hanoi Beijing Montreal London Paris Amsterdam Dublin Edinburgh Berlin Copenhagen Stockholm

This is a demo recommender system which can suggest ideas about best places to visit based on your previous travel experience combined with recommendations of other users.

How does it work?

Based on your current location, we create the map of up to 49 cities around the world you may have visited. If you don't find suitable destination, click “more locations” to get more suggestions. From the map select several spots you have visited and liked. In other words pick the destinations you can recommend to others. Click “recommend”.

The algorithm will find people with similar travel experience. For example say you selected Miami, Paris, Rome, Berlin. There is another user who visited Miami, Paris, Rome, Barcelona. The experience of you two are quite similar. So most likely you will be recommended Barcelona, which is a great city for your vacation by the way.

Technology behind

This application uses Apache Mahout as its recommender engine. It may help you learning how Mahout, collaborative filtering algorithm and recommender systems in general work in real examples.

Your feedback is appreciated

After getting your travel recommendation, you can rate it using thumbs up/down. This helps the algorithm to improve recommendations. We encourage to use that option as your feedback is more than welcome. Also you can browse through next recommender places to visit. It is a collective efforts to learn from travel experience of each other.

Partners

Airhint - flight price predictions with machine learning