soumajit11/insight-ml ? reverse-engineered prompt

Build me a simple Flask web app where I can upload a CSV or Excel file, preview the data safely, and get automatic analysis of the dataset. I want it to show basic stats like row and column counts, missing values, duplicates, column types, and a correlation view for numeric columns.

Then let me pick a target column and have the app figure out whether it is a classification or regression problem. After that, it should automatically clean and prepare the data, train a few common machine learning models, compare them, and show which one performed best. Please include a clear results page with metrics, model comparison, and feature importance if available.

I also want a way to export the trained model together with the preprocessing steps as a downloadable package, plus a CSV export of the results. Make the interface clean and easy to use, with a file upload area, analyze button, training controls, and download links. If you need to check current library docs online, feel free to do that.