Diffusion models aim to describe and predict the temporal evolution of dynamic phenomena, finding application not only in the adoption of innovative products and technologies, but also in the analysis of environmental and economic variables. In particular, this thesis seeks to examine and explore some key features of the Bass model and the GGM model, applying them to time series of CO₂ emissions. The main objective is to address and overcome several limitations that may arise in classical applications, focusing on the advantages offered by an extension based on nonlinear mixed-effects models (BMME). First, the opportunity to simultaneously analyze multiple national time series is investigated in order to obtain a global view of the phenomenon. The use of nonlinear mixed-effects models allows for providing an ensemble estimate by exploiting the hierarchical structure of the data, accounting at the same time for country-specific characteristics and for dynamics shared across countries. In addition, the thesis addresses a critical issue related to parameter estimation when the diffusion process has not yet reached its phase of maximum expansion, as is often the case for developing countries. In such situations, classical univariate approaches frequently lead to unstable estimates or difficulties in interpreting the parameters. The analysis shows how the mixed-effects framework helps to overcome these limitations: by constraining the shape parameters to follow a common distribution, it becomes possible to obtain more reliable and interpretable results even for series that are still in an exponential growth phase.
I modelli di diffusione mirano a descrivere e prevedere l’evoluzione temporale di fenomeni dinamici, trovando applicazione non solo nell'adozione di nuove tecnologie ma anche nell'analisi di variabili ambientali ed economiche. In particolare, in questa tesi si intende esaminare e approfondire le caratteristiche del modello di Bass e del modello GGM, applicandoli alle serie storiche delle emissioni di CO2. L'obiettivo principale è affrontare e correggere i limiti strutturali che emergono dalle applicazioni classiche, concentrandosi sui vantaggi offerti dall'approccio a effetti misti (BMME). In primo luogo, si indaga l'opportunità di prendere in esame contemporaneamente più serie di dati nazionali al fine di ottenere una visione globale del fenomeno, attraverso l'utilizzo dei modelli non lineari ad effetti misti. Questo approccio consente di fornire una stima d'insieme, sfruttando la struttura gerarchica dei dati per considerare simultaneamente la specificità del singolo paese e le dinamiche comuni a livello globale. Oltre a ciò, si affronta la problematica legata alla stima dei parametri che si incontra quando il processo non è ancora giunto alla fase di massima espansione, come nel caso dei paesi in via di sviluppo. In queste situazioni, l'utilizzo dei modelli univariati classici produce spesso risultati instabili o non significativi. L'analisi condotta dimostra come l'approccio misto permetta di superare tale ostacolo: vincolando i parametri di forma a una distribuzione comune, è possibile ottenere risultati più attendibili anche per quelle serie storiche ancora in fase di crescita esponenziale.
La dinamica delle emissioni di CO2 nei paesi del mondo: un approccio basato sul modello di Bass e sulle sue estensioni a effetti misti.
VISENTIN, MARTINA
2025/2026
Abstract
Diffusion models aim to describe and predict the temporal evolution of dynamic phenomena, finding application not only in the adoption of innovative products and technologies, but also in the analysis of environmental and economic variables. In particular, this thesis seeks to examine and explore some key features of the Bass model and the GGM model, applying them to time series of CO₂ emissions. The main objective is to address and overcome several limitations that may arise in classical applications, focusing on the advantages offered by an extension based on nonlinear mixed-effects models (BMME). First, the opportunity to simultaneously analyze multiple national time series is investigated in order to obtain a global view of the phenomenon. The use of nonlinear mixed-effects models allows for providing an ensemble estimate by exploiting the hierarchical structure of the data, accounting at the same time for country-specific characteristics and for dynamics shared across countries. In addition, the thesis addresses a critical issue related to parameter estimation when the diffusion process has not yet reached its phase of maximum expansion, as is often the case for developing countries. In such situations, classical univariate approaches frequently lead to unstable estimates or difficulties in interpreting the parameters. The analysis shows how the mixed-effects framework helps to overcome these limitations: by constraining the shape parameters to follow a common distribution, it becomes possible to obtain more reliable and interpretable results even for series that are still in an exponential growth phase.| File | Dimensione | Formato | |
|---|---|---|---|
|
Visentin_Martina.pdf
accesso aperto
Dimensione
1.55 MB
Formato
Adobe PDF
|
1.55 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
https://hdl.handle.net/20.500.12608/105791