Categorizing 8-K Filings: A Comparison of Machine Learning Models
| dc.contributor.author | Montgomery, Meg | |
| dc.date.accessioned | 2023-09-18T22:05:36Z | |
| dc.date.available | 2023-09-18T22:05:36Z | |
| dc.date.updated | 2023-09-18T22:05:36Z | |
| dc.description.abstract | In the United States, public companies are required to submit form 8-K filings to the SEC within four days of any material event occurring. The timely nature of these announcements makes them important resources for investors, industry professionals, and researchers alike. As part of an 8-K filing, one Item, Item 5.02, is of particular interest as it announces the arrival and departure of key executive officers and board members. Machine learning and classification algorithms have been applied most commonly to 10-K and 10-Q SEC filings, but 8-K filings have only received little research attention. Studying the contents of 8-K filings by applying and comparing advanced classification algorithms is therefore of outmost importance. This study uses a big dataset of 8-K textual filings and applies 11 classification models via two different experiments: (a) identifying the Item based on 8-K textual inputs, and (b) categorizing the type of executive departure event described in Item 5.02. The study compares four traditional classification algorithms combined with two feature extraction techniques, a recurrent neural network model (LSTM), and two transformer models (SmallBERT and GPT-2). | |
| dc.identifier.uri | https://repository.library.cofc.edu/handle/123456789/5546 | |
| dc.language.rfc3066 | en | |
| dc.title | Categorizing 8-K Filings: A Comparison of Machine Learning Models | |
| dspace.entity.type |