Showing 1–2 of 2 results for author: Trifirò, D

Machine Learning approach for Credit Scoring

Authors: A. R. Provenzano, D. Trifirò, A. Datteo, L. Giada, N. Jean, A. Riciputi, G. Le Pera, M. Spadaccino, L. Massaron, C. Nordio

Abstract: In this work we build a stack of machine learning models aimed at composing a state-of-the-art credit rating and default prediction system, obtaining excellent out-of-sample performances. Our approach is an excursion through the most recent ML / AI concepts, starting from natural language processes (NLP) applied to economic sectors' (textual) descriptions using embedding and autoencoders (AE), goi… ▽ More In this work we build a stack of machine learning models aimed at composing a state-of-the-art credit rating and default prediction system, obtaining excellent out-of-sample performances. Our approach is an excursion through the most recent ML / AI concepts, starting from natural language processes (NLP) applied to economic sectors' (textual) descriptions using embedding and autoencoders (AE), going through the classification of defaultable firms on the base of a wide range of economic features using gradient boosting machines (GBM) and calibrating their probabilities paying due attention to the treatment of unbalanced samples. Finally we assign credit ratings through genetic algorithms (differential evolution, DE). Model interpretability is achieved by implementing recent techniques such as SHAP and LIME, which explain predictions locally in features' space. △ Less

Submitted 20 July, 2020; originally announced August 2020.

Comments: 28 pages, 10 figures

MSC Class: 62M45; 68T01; 68T50; 91G40

An Artificial Intelligence approach to Shadow Rating

Authors: Angela Rita Provenzano, Daniele Trifirò, Nicola Jean, Giacomo Le Pera, Maurizio Spadaccino, Luca Massaron, Claudio Nordio

Abstract: We analyse the effectiveness of modern deep learning techniques in predicting credit ratings over a universe of thousands of global corporate entities obligations when compared to most popular, traditional machine-learning approaches such as linear models and tree-based classifiers. Our results show a adequate accuracy over different rating classes when applying categorical embeddings to artificia… ▽ More We analyse the effectiveness of modern deep learning techniques in predicting credit ratings over a universe of thousands of global corporate entities obligations when compared to most popular, traditional machine-learning approaches such as linear models and tree-based classifiers. Our results show a adequate accuracy over different rating classes when applying categorical embeddings to artificial neural networks (ANN) architectures. △ Less

Submitted 20 December, 2019; originally announced December 2019.

Comments: 18 pages, 10 figures, 6 tables

MSC Class: 62M45; 91G40