University of Twente Student Theses

Login

Model Selection

Perez Adroher, Arturo (2018) Model Selection.

[img] PDF
2MB
Abstract:This thesis is on the comparison of model selection algorithms. It builds a framework for relevant algorithms or criteria that have been proposed so far in literature and that lead to a model easy on interpretation and with high predictive power. The properties and performance of these model selection algorithms are studied and compared. We further propose redeveloped algorithms based on information criteria such as Akaike Information Criterion, Bayesian Information criterion, R2 and shrink- age parameters functions like Least Absolute Shrinkage and Selection Operator and Elastic Net to evaluate models’ predictive power and interpretation adapted to Rabobank requirements. These algorithms are programmed using open-source Python packages for numerical computation and statistical modelling, and tested with real data supplied by the company.
Item Type:Essay (Master)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:31 mathematics
Programme:Industrial Engineering and Management MSc (60029)
Link to this item:http://purl.utwente.nl/essays/76580
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page