The worlds of Artificial Intelligence and Machine Learning have shown a surge in popularity over the last years, tracing the route for a continuous stream of new ideas and applications. Some economic sectors clearly show a natural propensity towards the concept of digital transformation of their business. On the other hand, some other market areas, often renowned for their tendency towards more traditional and conservative approaches, exhibit a certain aversion for such paradigm shift. The aversion is partially justifiable in terms of ratio between expected costs and benefits, which frequently leads small firms to lose interest in those kinds of investments. With this premise, I decided to try and figure a suitable framework for the development of a business information system with integrated features of machine learning algorithms, designed to be applied in a particular industry that at first sight might have little interest in such approach: the restaurant industry. The dissertation starts from the blueprint of an algorithm for the generation of synthetic datasets with the goal of being able to produce a very large amount of records, that will serves as sketch on which the data structure itself is improved. In parallel to the description of the various components of the simulation algorithm, the dissertation provides a narrow introduction to related topics, such as database theory, coding, data exploration, graph theory and networks, as well as examples with applications in the context of a restaurant business. Moving forward, on the base of the proposed data structure / framework, I exhibit several applications of popular machine learning models. In particular, the reader will find exploitation of unsupervised machine learning, such as K-means clustering, Gaussian Mixture Models, Apriori Association Rule and Kernel Density Estimator. Furtherly, I introduce the possibility of using the generated clusters as features in a supervised model. The final goal of the dissertation is therefore to advance a base to work on for the development of a business model in the restaurant industry based on the active management of information generated by internal analysis.
L'interesse verso lo sviluppo di nuovi orizzonti per le applicazioni di Intelligenza Artificiale e Machine Learning ha visto un'impennata nel corso degli ultimi anni. Alcuni settori mostrano chiaramente una naturale propensione verso la trasformazione digitale del proprio business di riferimento. Per contro, altre aree del mercato, spesso rinomate per la tendenza alla conservazione delle tradizioni, manifestano una certa avversione verso tale cambiamento. Tale avversione è in parte giustificabile in termini di rapporto tra costi previsti e benefici attesi, che frequentemente allontana le piccole o medie imprese da progetti di investimento nel digitale. Con questa premessa, ho deciso di provare ad immaginare e progettare le linee guida per la realizzazione di un sistema informativo che possa far perno sull’uso di tecniche di machine learning, da applicare in un contesto che all'apparenza lascia poco spazio alla digitalizzazione: la ristorazione. Il lavoro parte dalla progettazione di un sistema per la generazione di dati sintetici - simulati - avente il fine di produrre un'elevata mole di informazioni, sulle quali immaginare la struttura di dati più adatta da applicare. Parallelamente alla descrizione dettagliata dell'algoritmo di simulazione, vengono forniti vari spunti di introduzione alla teoria delle basi di dati, programmazione logica, ed esplorazione dei dati, teoria dei grafi e network, il tutto corredato da esempi di applicazioni nel contesto della ristorazione. Infine, sulla base della struttura di dati proposta, vengono esposte alcune applicazioni di famosi modelli di machine learning. In particolare, vengono affrontate tecniche di apprendimento non supervisionato, come il clustering K-means e Gaussian Mixture Model, Apriori Association Rule e Kernel Density Estimator. Inoltre, viene fornito un particolare possibile studio dei dati mediante un algoritmo di apprendimento supervisionato, un albero decisionale per la regressione, costruito sulle informazioni individuate tramite clustering. Lo scopo finale dell'elaborato è quello di fornire una base di partenza per lo sviluppo di un modello di business nel settore della ristorazione basato sulla gestione attiva delle informazioni di analisi interna.
Applicazioni di Machine Learning per il Business Intelligence nel settore della ristorazione
BIANCO, MATTIA
2019/2020
Abstract
The worlds of Artificial Intelligence and Machine Learning have shown a surge in popularity over the last years, tracing the route for a continuous stream of new ideas and applications. Some economic sectors clearly show a natural propensity towards the concept of digital transformation of their business. On the other hand, some other market areas, often renowned for their tendency towards more traditional and conservative approaches, exhibit a certain aversion for such paradigm shift. The aversion is partially justifiable in terms of ratio between expected costs and benefits, which frequently leads small firms to lose interest in those kinds of investments. With this premise, I decided to try and figure a suitable framework for the development of a business information system with integrated features of machine learning algorithms, designed to be applied in a particular industry that at first sight might have little interest in such approach: the restaurant industry. The dissertation starts from the blueprint of an algorithm for the generation of synthetic datasets with the goal of being able to produce a very large amount of records, that will serves as sketch on which the data structure itself is improved. In parallel to the description of the various components of the simulation algorithm, the dissertation provides a narrow introduction to related topics, such as database theory, coding, data exploration, graph theory and networks, as well as examples with applications in the context of a restaurant business. Moving forward, on the base of the proposed data structure / framework, I exhibit several applications of popular machine learning models. In particular, the reader will find exploitation of unsupervised machine learning, such as K-means clustering, Gaussian Mixture Models, Apriori Association Rule and Kernel Density Estimator. Furtherly, I introduce the possibility of using the generated clusters as features in a supervised model. The final goal of the dissertation is therefore to advance a base to work on for the development of a business model in the restaurant industry based on the active management of information generated by internal analysis.È consentito all'utente scaricare e condividere i documenti disponibili a testo pieno in UNITESI UNIPV nel rispetto della licenza Creative Commons del tipo CC BY NC ND.
Per maggiori informazioni e per verifiche sull'eventuale disponibilità del file scrivere a: unitesi@unipv.it.
https://hdl.handle.net/20.500.14239/775