Profet-Style AutoML Dashboard: Visual Machine Learning Pipeline Generator

In this project, I designed and developed an advanced AutoML dashboard using Profet AI, allowing the user to quickly and visually experiment with thousands of model combinations and preprocessors.

Project goal:

Build an interactive platform that allows testing the impact of each step in the machine learning chain—from missing value handling and encoding to model selection—to understand the relationship between processing steps and final performance.

Main functions:

Automatically upload and analyze CSV files.

Choose or test all preprocessing techniques, such as:

? Missing values (Mean, Median, Most Frequent)

Encoding (Label, One-Hot)

Outliers (Z-score, IQR)

Balance (SMOTE, Under/Over Sampling)

Feature selection and dimensionality reduction (PCA, t-SNE)

Train more than 10 machine learning algorithms, such as XGBoost, SVM, and Random Forest.

Display results via an interactive dashboard including:

Performance charts (Accuracy, F1-Score)

️ Heatmaps for comparing models

Ranking table of best results

? Automatic PDF report of best models

Tools used:

Python – Streamlit – Scikit-learn – XGBoost – Pandas – Matplotlib – Plotly – SMOTE

The result:

An intelligent dashboard helps data analysts and researchers discover the best models and processes in minutes, with an elegant and easy-to-use interface.