Multi-modal deep learning for credit rating prediction using text and numerical data streams
Bibliographic record
Abstract
Knowing which factors are significant in credit rating assessments leads to better decision-making. However, the focus of the literature thus far has been mostly on structured data, and fewer studies have addressed unstructured or multimodal datasets. In this paper, we present an analysis of the most effective architectures for the fusion of deep learning models to predict company credit rating classes, using structured and unstructured datasets of different types. In these models, we tested various combinations of fusion strategies with selected deep-learning models, including convolutional neural networks (CNNs) and variants of recurrent neural networks (RNNs), and pre-trained language models (BERT). We study data fusion strategies in terms of level (including early and intermediate fusion) and techniques (including concatenation and cross-attention). Our results show that a CNN-based multi-modal model with a hybrid fusion strategy outperformed other multimodal techniques. In addition, by comparing simple architectures with more complex ones, we found that more sophisticated deep learning models do not necessarily produce the highest performance. Furthermore, we found that the text channel plays a more significant role than numeric data, with the contribution of text achieving an AUC of 0.91, while the maximum AUC of numeric channels was 0.808. Finally, rating agencies on short, medium, and long-term performance show that Moody’s credit ratings outperform those of other agencies like Standard & Poor’s and Fitch Ratings. • We investigate fusion strategies and deep-learning models in credit prediction. • We quantify the contribution of structured and unstructured data in the best model. • We explore COVID-19’s impact on model performance and its crisis adaptability. • We assess rating agencies’ performance over short, medium, and long terms.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.007 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".