The streamlined version of the mean field variational Bayes (MFVB) algorithm for linear mixed models with crossed random effects allows simplifying calculations but may require one group’s dimension to be moderate. Data collecting high school students’ first term evaluations and INVALSI scores for Italian and Maths subjects perfectly comply with this setting: students are a vast random sample of those who enrolled at the university in 2019/20, while the number of tests is limited to 6. Three different MFVB product restrictions with incremental complexity are evaluated. All of them are convenient with respect to classic MCMC solutions from both a computational and a memory storage viewpoint. Results and interpretation of model coefficients are in line with the literature on educational data

Streamlined Variational Inference for Modeling Italian Educational Data

Gioia Di Credico;Claudia Di Caterina;Francesco Santelli
2023-01-01

Abstract

The streamlined version of the mean field variational Bayes (MFVB) algorithm for linear mixed models with crossed random effects allows simplifying calculations but may require one group’s dimension to be moderate. Data collecting high school students’ first term evaluations and INVALSI scores for Italian and Maths subjects perfectly comply with this setting: students are a vast random sample of those who enrolled at the university in 2019/20, while the number of tests is limited to 6. Three different MFVB product restrictions with incremental complexity are evaluated. All of them are convenient with respect to classic MCMC solutions from both a computational and a memory storage viewpoint. Results and interpretation of model coefficients are in line with the literature on educational data
2023
978-88-6952-170-6
https://www.paviauniversitypress.it/catalogo/proceedings-of-the-statistics-and-data-science-conference/6705
http://archivio.paviauniversitypress.it/oa/9788869521706.pdf
File in questo prodotto:
File Dimensione Formato  
DiCredicoDiCaterinaSantelliPAVIA.pdf

accesso aperto

Descrizione: Copertina, indice e articolo
Tipologia: Documento in Versione Editoriale
Licenza: Creative commons
Dimensione 7.93 MB
Formato Adobe PDF
7.93 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11368/3050721
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact