This project aims to analyze financial news articles and predict their sentiment using a pre-trained sentiment analysis model. The sentiment analysis focuses on financial news to help understand market trends and investor sentiment, which can be crucial for decision-making in stock trading and investments.
- Sentiment Analysis: Predicts whether financial news articles have a positive, neutral, or negative sentiment.
- Model Integration: Utilizes a fine-tuned version of
distilrobertaoptimized for financial sentiment analysis. - Dataset Support: Processes financial datasets for model training and evaluation.
We use the DistilRoBERTa model fine-tuned for financial sentiment analysis, published on Hugging Face by mrm8488.
Model link: DistilRoBERTa Fine-tuned Financial News Sentiment Analysis
This model is specifically fine-tuned to analyze financial texts, making it highly suitable for extracting sentiments from financial news articles, unlike generic sentiment analyzers.
The project uses financial news datasets made available on Hugging Face by NickyNicky.
Dataset link: Financial News Dataset
- Comprises financial news articles tagged with sentiment labels.
- Suitable for both model training and evaluation.
- Loss: 0.4090
- Accuracy: 0.9171
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- num_epochs: 5
| Training Loss | Epoch | Validation Loss |
|---|---|---|
| 0.318500 | 1.0 | 0.294045 |
| 0.281700 | 2.0 | 0.298364 |
| 0.250100 | 3.0 | 0.302255 |
| 0.186400 | 4.0 | 0.380530 |
| 0.179100 | 5.0 | 0.409072 |