MLChatAgent

What it does

The app is designed to help users interact with complex machine learning (ML) models effortlessly. It simplifies understanding and working with ML models by allowing users to ask questions and receive detailed, comprehensible answers in English.

Using the Gemini chat agent, the app employs several tools to address user questions:

- question_reformer: Reformulates questions for clarity.
- generate_sql: Generates SQL queries
- execute_sql: Executes queries and summarizes data.
- subset_churn_contribution_analysis: Performs churn contribution analysis on subsets
- subset_clv_analysis: Evaluates CLV impact based on treatments done on subsets selected.
- subset_shap_summary: Provides insights behind predictions and next best actions.
- customer_recommendations: Suggests ways to reduce churn for individuals.
- model_stat: Answers model-related questions.
- generate_visualizations: Creates visual data representations.

All these tools are individual agents powered by Gemini models

Here's how the app works:

- User Interaction: Users ask questions in plain English.
- Planning: The agent identifies necessary tools and plans the execution.
- Translation: The agent reframes questions into concise instructions.
- Model Interaction: The agent executes the plan using ML models.
- User Response: The agent interprets and delivers responses in plain English.

This agentic framework empowers business users to fully leverage ML models and data without needing highly technical skills.