This thesis presents the activities carried out during an internship at Venis S.p.A., the company that designs, develops and manages the City of Venice's IT System and Telecommunications network. The work focuses on two main projects, while briefly mentioning the additional activities that were carried out during the experience. The first project focused principally on the development of a predictive model to estimate the number of people expected to be present in the city of Venice during the Carnival period. The project included data extraction, preprocessing, feature selection and model evaluation. Finally, several predictive models were compared to identify the most suitable one to adopt in future decision-making in the management of urban events. The second project concerned the design and implementation of an AI-powered chatbot based on the OpenAI ChatGPT API. It was built to provide users with accurate and context-aware responses on different topics. The main focus was on the IMU (Municipal property tax), with the future idea of expanding the project to all municipal tax services. The experience provided a valuable opportunity to gain practical experience in predictive analytics, data processing, the use of real-world data and large language model integration. It also enabled the application of the theoretical knowledge acquired during the Master's degree in Data Science.

Predictive analytics and Generative AI for Public Services

MARIETHOZ, JEAN-ZACHARIE
2025/2026

Abstract

This thesis presents the activities carried out during an internship at Venis S.p.A., the company that designs, develops and manages the City of Venice's IT System and Telecommunications network. The work focuses on two main projects, while briefly mentioning the additional activities that were carried out during the experience. The first project focused principally on the development of a predictive model to estimate the number of people expected to be present in the city of Venice during the Carnival period. The project included data extraction, preprocessing, feature selection and model evaluation. Finally, several predictive models were compared to identify the most suitable one to adopt in future decision-making in the management of urban events. The second project concerned the design and implementation of an AI-powered chatbot based on the OpenAI ChatGPT API. It was built to provide users with accurate and context-aware responses on different topics. The main focus was on the IMU (Municipal property tax), with the future idea of expanding the project to all municipal tax services. The experience provided a valuable opportunity to gain practical experience in predictive analytics, data processing, the use of real-world data and large language model integration. It also enabled the application of the theoretical knowledge acquired during the Master's degree in Data Science.
2025
Predictive analytics and Generative AI for Public Services
Predictive model
LLM
Real-world data
Public services
File in questo prodotto:
File Dimensione Formato  
Mariethoz_JeanZacharie_thesis.pdf

accesso aperto

Dimensione 2.96 MB
Formato Adobe PDF
2.96 MB Adobe PDF Visualizza/Apri

The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12608/115844