How soumajit11/insight-ml works
This is a Flask-based AutoML web application. Users upload a CSV or Excel dataset, inspect a secure preview and automated EDA, configure a target and split, compare trained models, then review metrics and feature importance or export results and a model package. The graph reflects documented capabilities and sampled call paths; some route and export wiring is described in the README but not visible in the supplied source excerpts.
Main entry points: model_utils.py ML facade · app.py Flask routes · app_helpers.py Request and session helpers · app.js Browser UI
Components 17
Web experience
- Browser UI
static/app.js - Page templates
templates - Flask routes
app.py
Dataset workflow
- Request and session helpers
app_helpers.py - Uploaded dataset
- Session cache
app_helpers.py - Safe preview rendering
ml_utils/config.py
Analysis and training
- ML facade
model_utils.py - Validation and target detection
ml_utils/utils.py - Exploratory analysis
ml_utils/eda.py - Training configuration
ml_utils/config.py - Feature preprocessing
ml_utils/preprocessing.py - Model comparison
ml_utils/models.py
Results and export
- Metrics and importance
ml_utils/models.py - Model and results export
app.py - Downloadable model package
Other components
- User
Connections 21
- User → Browser UI: uses
- Page templates → Browser UI: renders
- Browser UI → Flask routes: sends requests
- Flask routes → Request and session helpers: dispatches
- Request and session helpers → Uploaded dataset: validates and loads
- Request and session helpers → Session cache: caches session data
- Request and session helpers → ML facade: requests analysis and training
- Flask routes → ML facade: requests target analysis
- ML facade → Validation and target detection: validates and detects
- ML facade → Exploratory analysis: computes EDA
- ML facade → Safe preview rendering: builds preview
- ML facade → Training configuration: creates training config
- ML facade → Model comparison: runs comparison
- Model comparison → Feature preprocessing: uses pipelines
- Model comparison → Metrics and importance: produces
- Request and session helpers → Session cache: stores training results
- Flask routes → Session cache: reads cached results
- Flask routes → Model and results export: handles export requests
- Model and results export → Downloadable model package: streams model package
- Model and results export → Browser UI: returns downloads
- Flask routes → Browser UI: returns API results