Disseny d'un algorisme per a l'assignació de paquets de viatge
Correu electrònic de l'autoralex.ruaestudiant.upc.edu
Tutor / director / avaluadorLópez Masip, Susana Clara
Tipus de documentTreball Final de Grau
Condicions d'accésAccés restringit per acord de confidencialitat
The project has been developed in a company that offers destination surprise travels. Travellers have a possible destinations list, but they do not know where they are travelling to until two days before the departure flight. The purpose of this document is to develop a tool that reduces the workload of one of the biggest departments of the company. With this tool, it is possible to speed up the destination and hotel choosing processes, at the same time than increasing the company profit and the customers satisfaction. On the one hand, it has been observed that depending on the origin, in average, customer feedbacks are in some destinations different, but not the accomodation ones. On the other hand, it has been concluded that destinations with bad feedback or places where people did not want to travel to, were very few and this parameter did not have a big effect on overall punctuation. However, the hotel's location as well as its quality have a bigger influence in customers' feedback. The computers' speed and their ability to deal with large amount of data, make it possible to reach results impossible to get using only traditional staff's work. In order to carry on with this technique, it is necessary to search for places to find the data and then to load it with computer software, which simulates a crowd surfing the internet. It means a robot making petitions to servers. Once this amount of data is obtained, it is necessary to find methods; first to save them all and then to manage them. In this case, data bases have been used. Data has been split by categories and these categories have been combined between them resulting in an even larger number of possibilities and results. Filters have been used to reduce the number of combinations to those more likely and significant, discarding the other ones. Last but not least, a method used to use artificial intelligence, have self learning and to check mistakes is to keep feedbacks uploaded to the data base. Consequently, in case the customers do not like a destination or a hotel, the algorithm, using a threshold value, will discard them. At the end of the project, it has been proved that the profit per client as well as their satisfaction has considerably increased.