The thesis proposes a data-driven framework for automating the planning of training offerings within an Italian provider of professional training. The work is structured along two complementary levels: an organizational analysis of decision-making processes and the quantitative exploitation of company data. The first part consists of a systematic mapping of operational and information flows across the Training, Technical, and Marketing departments, identifying inefficiencies and automation opportunities in current decision-making processes. The second part leverages the company’s databases to perform unsupervised segmentation of the course portfolio through clustering algorithms. The resulting clusters serve as the foundation for developing cluster-specific predictive models aimed at forecasting the number of course editions to be scheduled three months in ahead and their temporal distribution. The approach integrates qualitative process analysis methods with machine learning techniques (clustering and time-series forecasting), contributing to the literature on digital transformation in the education sector while providing the partner company with a practical tool to reduce over- and under-planning risks, improve class fill rates, and enhance overall operational efficiency. The expected outcome is a prototype of an automated planning system.
La tesi propone un framework data-driven per l’automazione della pianificazione dell’offerta formativa in un provider italiano di formazione professionale. Il lavoro si sviluppa su due livelli complementari: l’analisi organizzativa dei processi decisionali e lo sfruttamento quantitativo dei dati aziendali. Nella prima parte viene condotta una mappatura sistematica dei flussi operativi e informativi dei dipartimenti Formazione, Tecnico e Marketing, evidenziando potenziali inefficienze e possibilità di automazione. Nella seconda parte, a partire dai database forniti, si realizza una segmentazione non supervisionata del portafoglio corsi mediante algoritmi di clustering. I cluster individuati vengono poi utilizzati come base per lo sviluppo di modelli predittivi specifici, finalizzati alla stima del numero di edizioni da programmare a tre mesi di distanza e alla loro distribuzione. L’approccio integra metodi di process analysis qualitativa con tecniche di machine learning (clustering e forecasting su serie temporali), contribuendo alla letteratura sulla trasformazione digitale nel settore della formazione e offrendo all’azienda uno strumento per ridurre il rischio di over-/under-planning, migliorare il tasso di riempimento delle classi e incrementare l’efficienza operativa complessiva. Il risultato atteso è un prototipo di sistema di pianificazione automatica.
Data-Driven Automation of Course Planning: Portfolio Clustering and Disaggregated Demand Forecasting in a Professional Training Provider
FACINI, ANTONELLA
2024/2025
Abstract
The thesis proposes a data-driven framework for automating the planning of training offerings within an Italian provider of professional training. The work is structured along two complementary levels: an organizational analysis of decision-making processes and the quantitative exploitation of company data. The first part consists of a systematic mapping of operational and information flows across the Training, Technical, and Marketing departments, identifying inefficiencies and automation opportunities in current decision-making processes. The second part leverages the company’s databases to perform unsupervised segmentation of the course portfolio through clustering algorithms. The resulting clusters serve as the foundation for developing cluster-specific predictive models aimed at forecasting the number of course editions to be scheduled three months in ahead and their temporal distribution. The approach integrates qualitative process analysis methods with machine learning techniques (clustering and time-series forecasting), contributing to the literature on digital transformation in the education sector while providing the partner company with a practical tool to reduce over- and under-planning risks, improve class fill rates, and enhance overall operational efficiency. The expected outcome is a prototype of an automated planning system.| File | Dimensione | Formato | |
|---|---|---|---|
|
900455_FACINI.pdf
non disponibili
Dimensione
2.88 MB
Formato
Adobe PDF
|
2.88 MB | Adobe PDF |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14247/27946