BA882: Predictive End-to-End Analytics Pipeline for Financial APIs Group 6

Project Overview: Predictive Financial Analytics APIs

The Predictive Financial Analytics APIs project is a comprehensive end-to-end solution designed to provide actionable insights into daily stock data for selected technology companies. This pipeline integrates multiple components to deliver a robust financial analysis toolset:

Pipeline Features:

1. Daily Stock Data Acquisition:
   The pipeline automatically extracts, processes, and updates daily stock data for key technology companies (e.g., Apple, Microsoft, Nvidia, Amazon, Netflix).

2. News Integration and Sentiment Analysis:

   • Aggregates the latest news articles related to the selected companies.
   • Utilizes sentiment analysis to score news data, providing insights into market sentiment for each company.

3. Predictive Modeling:

   • Implements traditional machine learning models like XGBoost Regressor and Random Forest for reliable predictions.
   • Employs advanced neural network models, including LSTM (Long Short-Term Memory) and RNN (Recurrent Neural Networks), to predict key financial metrics such as:
     • Closing Prices: Forecasts daily stock closing prices.
     • Trade Volumes: Predicts trade volumes with high accuracy.

4. Interactive User Interface:
   A Streamlit-based application serves as the user interface, offering:

   • BigQuery Integration: Enables users to text-to-sql query real-time and historical data from BigQuery.
   • Large Language Model (LLM) Integration: Allows users to interact with financial data via natural language queries.
   • News Insights: Displays the latest news articles alongside sentiment scores for easy interpretation.

5. 10-K Document Analysis:

   • Users can upload 10-K reports of the selected companies directly into the interface.
   • An integrated LLM-based chatbot analyzes the uploaded documents, enabling users to ask specific questions about the financial conditions and performance metrics of each company.

Key Highlights:

• Automation: Fully automated data pipeline orchestrates stock data extraction, transformation, and loading (ETL), coupled with news updates.
• Machine Learning and AI: Combines traditional and neural network-based predictive models for enhanced accuracy.
• User-Centric Design: The Streamlit application ensures accessibility, making it easy for users to obtain predictions, analyze sentiment, and engage with financial reports interactively.
• Comprehensive Insights: By integrating financial metrics, predictive analytics, sentiment analysis, and document review, the project delivers a holistic view of each company’s financial health and market outlook.

Sources and Tools Utilized in the Project

Data Source

• Alpha Vantage
  • Website: Alpha Vantage
  • Documentation: API Documentation

Technology Stocks Used

1. Apple (AAPL)
2. Microsoft (MSFT)
3. Nvidia (NVDA)
4. Netflix (NFLX)
5. Amazon (AMZN)

Tools

1. Google Cloud
2. Prefect Orchestration & Prefect Cloud
3. Google BigQuery
4. Flask
5. Google Secret Manager
6. Pinecone
7. Langchain
8. Vertex AI
9. Streamlit

Streamlit User-Interface:

• https://streamlit-genai-apps-676257416424.us-central1.run.app

Looker Visualization:

• https://lookerstudio.google.com/s/kDvwmHdQ76g