Multicriteria interpretability driven deep learning

Repetto, Marco

doi:10.1007/s10479-022-04692-6

Cited by 10 publications

(4 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The SHAP method was introduced to test the predictive power of each variable, highlighting the core features in predicting the financial sector stock market price features. In addition to SHAP, this framework provides powerful interpretability of model performance (Repetto, 2022 ). In fact, understanding the sources and processes of the predictive power of COVID-19-related government interventions for financial sector stock market returns and volatility can help support financial markets.…”

Section: Discussionmentioning

confidence: 99%

“…On the one hand, government interventions signal changes in future economic conditions, which rationally restructure asset portfolios both within and across asset classes (Zaremba et al, 2020 ). On the other hand, government interventions imply a potential divergence of opinions, which could increase irrational behaviors (Repetto, 2022 ). Government interventions contribute to fluctuations in market economies that are important for forecasting industry stock markets.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An interpretable system for predicting the impact of COVID-19 government interventions on stock market sectors

et al. 2023

View full text Add to dashboard Cite

Evaluating and understanding the financial impacts of COVID-19 has emerged as an urgent research agenda. Nevertheless, the impacts of government interventions on stock markets remain poorly understood. This study explores, for the first time, the impact of COVID-19 related government intervention policies on different stock market sectors using explainable machine learning-based prediction models. The empirical findings suggest that the LightGBM model provides excellent prediction accuracy while preserving computationally efficient and easy explainability of the model. We also find that COVID-19 government interventions are better predictors of stock market volatility than stock market returns. We further show that the observed effects of government intervention on the volatility and returns of ten stock market sectors are heterogeneous and asymmetrical. Our findings have important implications for policymakers and investors in terms of promoting balance and sustaining prosperity across industry sectors through government interventions.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

An interpretable system for predicting the impact of COVID-19 government interventions on stock market sectors

et al. 2023

View full text Add to dashboard Cite

show abstract

“…The first is given by the post-hoc nature of ALEs: post-hoc methods restrict the possibility to address any biases and impose some sort of regularization on the interpretations. 86 On the user's side, they require some basic knowledge of the methodology to interpret its outcomes. Second, the cross-sectional nature of the data prevented us from including in the analysis standard non-firm specific predictors, such as regional GDP growth, industry-level value added or business confidence indicators, which could have helped to reduce classification errors.…”

Section: Discussionmentioning

confidence: 99%

“…This study has some limitations revolving around three main aspects. The first is given by the post‐hoc nature of ALEs: post‐hoc methods restrict the possibility to address any biases and impose some sort of regularization on the interpretations 86 . On the user's side, they require some basic knowledge of the methodology to interpret its outcomes.…”

Section: Discussionmentioning

confidence: 99%

Lost in a black‐box? Interpretable machine learning for assessing Italian SMEs default

Crosato,

Liberati,

Repetto

2023

Appl Stoch Models Bus & Ind

View full text Add to dashboard Cite

Academic research and the financial industry have recently shown great interest in Machine Learning algorithms capable of solving complex learning tasks, although in the field of firms' default prediction the lack of interpretability has prevented an extensive adoption of the black‐box type of models. In order to overcome this drawback and maintain the high performances of black‐boxes, this paper has chosen a model‐agnostic approach. Accumulated Local Effects and Shapley values are used to shape the predictors' impact on the likelihood of default and rank them according to their contribution to the model outcome. Prediction is achieved by two Machine Learning algorithms (eXtreme Gradient Boosting and FeedForward Neural Networks) compared with three standard discriminant models. Results show that our analysis of the Italian Small and Medium Enterprises manufacturing industry benefits from the overall highest classification power by the eXtreme Gradient Boosting algorithm still maintaining a rich interpretation framework to support decisions.

show abstract

Forecasting corporate financial performance with deep learning and interpretable ALE method: Evidence from China

Liang,

Liu,

et al. 2024

Journal of Forecasting

View full text Add to dashboard Cite

Forecasting and analyzing corporate financial performance are of significant value to investors, managers, and regulators. In this paper, we constructed the one‐dimensional convolutional neural networks (1D‐CNN) and long short‐term memory (LSTM) deep learning models to investigate the feasibility of forecasting corporate financial performance with deep learning models, using the corporate financial features and environment, social and governance (ESG) rating index of Chinese A‐share listed corporation data from 2015 to 2021. Five evaluation metrics were employed to measure models' forecasting effects, and four competing machine learning models were built to verify the improvement in forecasting accuracy brought by the deep learning models. Furthermore, we also introduced the Accumulated Local Effects method to explore the forecasting processes of the deep learning models. The empirical results show the following: (1) Deep learning models can effectively extract the time‐series information in corporate data, thereby solving the task of predicting corporate financial performance with high accuracy. (2) The introduction of ESG information significantly contributes to the forecasting accuracy of corporate financial performance. For both 1D‐CNN and LSTM models, the ESG rating index can provide additional useful information for forecasting. (3) The interpretable results reveal the preference and emphasis of the two deep learning models for the different features. This further proves the robustness and reliability of deep learning models in forecasting corporate financial performance.

show abstract

Multicriteria interpretability driven deep learning

Cited by 10 publications

References 39 publications

An interpretable system for predicting the impact of COVID-19 government interventions on stock market sectors

An interpretable system for predicting the impact of COVID-19 government interventions on stock market sectors

Lost in a black‐box? Interpretable machine learning for assessing Italian SMEs default

Forecasting corporate financial performance with deep learning and interpretable ALE method: Evidence from China

Contact Info

Product

Resources

About