Napovedovanje časa čiščenja hotelskih sob na podlagi značilnosti gostov

Oblak, Gal (author), Vavpotič, Damjan (mentor)

Podatkovna znanost je zaradi vedno večje količine vsakodnevno nastalih podatkov vedno bolj aktualna. Z diplomsko nalogo smo želeli pomagati podjetju, ki razvija računalniško rešitev za podporo hotelskih procesov in pri tem zbira vrsto podatkov, ki nastajajo v okviru izvajanja le teh. Za boljše razumevanje področja so najprej opisane metode, ki smo jih pri napovedovanju uporabili. Nato smo opisali pristop napovedovanja dolžine čiščenja sob ter izgradnje napovednih modelov. Z analizo podatkov podjetja in izdelavo različnih napovednih modelov smo v okviru diplomske naloge želeli odkriti profile gostov, za katerimi je potrebno daljše čiščenje, ter tiste, za katerimi zadošča krajše čiščenje sobe. Cilj naloge je bila tudi opredelitev spremenljivk, ki dejansko vplivajo na dolžino čiščenja. Rezultati analize so pokazali, da bi za izdelavo bolj točnih napovednih modelov potrebovali še druge spremenljivke, vezane predvsem na hotel, potrebno pa bi bilo imeti tudi podatke za več različnih hotelov. Napovedni modeli so bili izdelani za posamezne hotele, saj so bile razlike v času čiščenja sob med različnimi hoteli prevelike, da bi lahko izdelali splošen napovedni model. Rezultati podjetju pomagajo pri razumevanju dejavnikov, ki vplivajo na čas čiščenja.

• Language: Slovenian
• Keywords: podatkovno rudarjenje, analiza podatkov, podatkovna znanost, napovedni modeli, vizualizacija podatkov
• Work type: Bachelor thesis/paper
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2018
• PID: 20.500.12556/RUL-103078
• Publication date in RUL: 13.09.2018

Secondary language

• Language: English
• Title: Predicting the cleaning time of hotel rooms based on guests characteristics

Abstract

Data science is becoming more and more relevant due to the increasing amount of data generated every day. The aim of our Bachelor’s thesis was to assist a company that offers computer solutions to support hotel processes and collects series of data generated during these processes. Firstly, the methods that are used in predicting are presented. Then, the approach of predicting the duration of room cleaning and the construction of predictive models is defined. The goal of our thesis was to identify the profiles of guests that effect longer or shorter room cleaning time by using data analysis and producing various predictive models. Another goal was to define variables that actually affect the length of the cleaning. The results of the analysis showed that for the development of more precise forecasting models, more variables related primarily to the hotel are needed, but also having data of several different hotels is crucial. Created models were designed for individual hotel due to large differences among the time needed for room cleaning in different hotels. The results contributed to a better understanding of factors that influence the time of cleaning.

• Keywords: data mining, data analysis, data science, predictive models, data visualization