The following work focuses on exploring the untapped potential of traditional machine learning methods to create a support system for a company in charge of civil air traffic management in Italy. It aims to provide employees and managers with tools that can harness functions and information from various central systems and orchestrate them in a way that simplifies business processes and improves the employee journey through the company's systems. By training both traditional and innovative models, accurate predictions of employees' absences for various time intervals will be obtained, preparing for a subsequent integration within an existing platform developed to support the company's systems. The problem is modelled as a classification problem, extracting meaningful time intervals that will be exploited to estimate the absence classes. The performance of the models was evaluated using state-of-the-art metrics for each analyzed case. It will also be demonstrated how the deployed methods can be applied to other workforce management and human resources management cases. Through its experimental nature, this thesis wants to provide organisations with innovative tools to anticipate and manage employee absenteeism. By encouraging the exploration of models adapted to specific industry contexts and the integration of advanced techniques to continuously improve forecasting accuracy, the research lays the foundation for further experimentation.
Il seguente lavoro si concentra sull'esplorazione del potenziale non sfruttato del machine learning per creare un sistema a supporto di una società incaricata della gestione del traffico aereo civile in Italia. Il suo obiettivo è quello di fornire ai dipendenti e ai manager strumenti in grado di sfruttare le funzioni e le informazioni provenienti da diversi sistemi centrali e di orchestrarli in modo da semplificare i processi aziendali e migliorare il workflow. Grazie all'applicazione di modelli di machine learning, sia tradizionali che innovativi, verranno effettuate previsioni accurate sulle assenze degli impiegati nei vari intervalli di tempo, per una successiva integrazione all'interno di una piattaforma già esistente sviluppata a supporto dei sistemi dell'azienda. Il problema viene modellato come un problema di classificazione, estraendo intervalli significativi sui quali verrà effettuata la stima delle classi di assenza. Le performance dei modelli sono state valutate usando metriche specifiche per ogni caso in analisi. Si dimostrerà inoltre come tale analisi possa essere applicata ad altre casistiche legate alla gestione delle risorse umane. Attraverso la sua natura sperimentale, questa tesi mira a fornire alle organizzazioni strumenti innovativi per anticipare e gestire le assenze dei dipendenti. Incoraggiando l'esplorazione di modelli adattati a contesti industriali specifici e l'integrazione di tecniche avanzate per migliorare continuamente l'accuratezza della previsione, la ricerca pone le basi per ulteriori sperimentazioni.
Predictive absence modeling in air traffic control centers
Frangipane, Gianmarco
2023/2024
Abstract
The following work focuses on exploring the untapped potential of traditional machine learning methods to create a support system for a company in charge of civil air traffic management in Italy. It aims to provide employees and managers with tools that can harness functions and information from various central systems and orchestrate them in a way that simplifies business processes and improves the employee journey through the company's systems. By training both traditional and innovative models, accurate predictions of employees' absences for various time intervals will be obtained, preparing for a subsequent integration within an existing platform developed to support the company's systems. The problem is modelled as a classification problem, extracting meaningful time intervals that will be exploited to estimate the absence classes. The performance of the models was evaluated using state-of-the-art metrics for each analyzed case. It will also be demonstrated how the deployed methods can be applied to other workforce management and human resources management cases. Through its experimental nature, this thesis wants to provide organisations with innovative tools to anticipate and manage employee absenteeism. By encouraging the exploration of models adapted to specific industry contexts and the integration of advanced techniques to continuously improve forecasting accuracy, the research lays the foundation for further experimentation.File | Dimensione | Formato | |
---|---|---|---|
2024_04_Frangipane.pdf
non accessibile
Descrizione: The following work focuses on exploring the untapped potential of traditional machine learning methods to create a support system for a company in charge of civil air traffic management in Italy
Dimensione
26.87 MB
Formato
Adobe PDF
|
26.87 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/217822