How luialv2502/ml_framework works
Laboratorio IA is a local Streamlit application for exploring datasets and running EDA, preprocessing, clustering, classification, regression, and multi-agent analyses. The user loads data, explores and prepares it, then runs an analysis and can review or compare saved experiments. The documented controller is the frontend’s sole entry point and delegates tasks to an agent coordinator; execution details for several unsampled agents and pages are represented only at subsystem level. Optional Gemini support and agent messaging are shown separately where evidence permits.
Main entry points: agents Analysis agents · controller.py Framework controller · app.py Streamlit entry · dashboard.py Dashboard
Components 27
Streamlit interface
- Streamlit entry
app.py - Dashboard
frontend/dashboard.py - Analysis pages
frontend/pages - Shared UI
frontend/components/shared.py - Supervised reports
frontend/components/supervised_agent.py - Visual theme
frontend/components/theme.py
Task orchestration
- Framework controller
framework_ia/controller.py - Agent agency
framework_ia/agents/agency.py - Task coordinator
framework_ia/agents/coordinator_agent.py - Agent runtime
framework_ia/agents/base_agent.py - Agent factory
framework_ia/agents/agent_factory.py - Agent messaging
framework_ia/agents/messaging.py - Gemini integration
framework_ia/agents/llm.py - Analysis agents
framework_ia/agents
Data analysis
- EDA operations
framework_ia/ml/eda/dataframe.py - Preprocessing
framework_ia/services/preprocessing_service.py - Data transforms
framework_ia/ml/preprocessing/transforms.py - Clustering and PCA
framework_ia/ml/clustering - Classification
framework_ia/ml/classification - Regression
framework_ia/ml/regression/regression.py
Data and persistence
- Dataset service
framework_ia/services/dataset_service.py - Experiment service
framework_ia/services/experiment_service.py - Experiment records
framework_ia/persistence/experiment_store.py
Shared infrastructure
- Input validation
framework_ia/utils/validation.py - Result serialization
framework_ia/utils/serialization.py
Other components
- Analyst
- Gemini
Connections 24
- Analyst → Streamlit entry: opens
README.md - Streamlit entry → Dashboard: starts
app.py - Dashboard → Visual theme: applies
frontend/dashboard.py - Analyst → Analysis pages: uses
README.md - Analysis pages → Shared UI: shares controls
frontend/pages/classification.py - Shared UI → Framework controller: calls
frontend/components/shared.py - Supervised reports → Visual theme: renders with
frontend/components/supervised_agent.py - Framework controller → Dataset service: loads active data
framework_ia/controller.py - Framework controller → Agent agency: dispatches task
framework_ia/controller.py - Agent agency → Task coordinator: routes task
- Task coordinator → Analysis agents: delegates analysis
- Agent factory → Agent runtime: constructs
framework_ia/agents/agent_factory.py - Agent runtime → Agent messaging: uses bus
framework_ia/agents/base_agent.py - Analysis agents → EDA operations: uses EDA
framework_ia/agents/eda_agent.py - Analysis agents → Preprocessing: uses preparation
framework_ia/agents/eda_agent.py - Preprocessing → Data transforms: applies operations
framework_ia/services/preprocessing_service.py - Analysis agents → Clustering and PCA: runs clustering
- Analysis agents → Classification: runs classification
framework_ia/agents/plugins/knn_auto.py - Analysis agents → Regression: runs regression
- Framework controller → Experiment service: records runs
framework_ia/controller.py - Experiment service → Experiment records: persists
framework_ia/services/experiment_service.py - Framework controller → Input validation: validates requests
framework_ia/controller.py - Agent messaging → Result serialization: serializes messages
framework_ia/agents/messaging.py - Gemini integration → Gemini: calls provider