59 lines (40 loc) · 1.93 KB

Financial Sentiment Analysis API

An API built with FastAPI for text analysis and sentiment detection, focusing on financial-related content.

Features

Sentiment Analysis: Returns sentiment (positive, negative, or neutral) based on the provided financial text.
Machine Learning Models: Integrates algorithms such as SVM, Naive Bayes, XGBoost, LightGBM, and others.
Text Preprocessing: Cleans and normalizes text before analysis.

Technologies Used

Key technologies and libraries used in this project include:

FastAPI: Modern, high-performance backend framework.
Pydantic: For data validation and serialization.
Scikit-learn: Tools for machine learning and data mining.
Joblib: For serialization of trained models.
NLTK: Tools for natural language processing.
PostgreSQL: For storing analysis data, now hosted on Neon.

API Endpoints

Interactive Documentation Access:

Swagger UI:
ReDoc:

Endpoints Overview

1. `POST /model_prediction`

Performs sentiment analysis based on a provided text and specified model.

2. `POST /multi_model_prediction`

Performs sentiment analysis using multiple models on a provided text.

Available Models:
- Naive Bayes
- SVM
- XGBoost
- LightGBM
- Multilayer Perceptron
- Gradient Boosting
- Random Forest
- AdaBoost
- Decision Tree

About the Models

Access the repository for more details:

Financial Market Sentiment Prediction: https://github.com/erickmaiia/ml-financial-sentiment-analysis
Download the BERT model here and place it in the app/ml_models/BERT directory: https://drive.google.com/file/d/1YkO0rkNpEkX91WFS2VbNgjJC_Z0qOn15/view?usp=sharing