← Back to portfolio

AUTOML · XGBOOST · MODEL EVALUATION

AutoML and XGBoost Model Selection

A graduate machine-learning project using automated model search to compare candidate pipelines and identify a strong classification model.

Recreated AutoML and XGBoost model-selection dashboard
Portfolio visualization recreated with sample or illustrative data to demonstrate the project’s analytical approach.

OBJECTIVE

Determine whether an automated machine-learning workflow could efficiently identify a competitive predictive pipeline while maintaining clear performance evaluation.

APPROACH

Configured an AutoML experiment, evaluated candidate preprocessing and model combinations, compared classification performance, and selected an XGBoost-based solution that achieved an AUC of approximately 0.79.

ANALYTICAL VALUE

Demonstrated efficient model experimentation, disciplined metric comparison, and the ability to move from automated search results to a defensible model recommendation.

SKILLS SHOWN

AutoML experimentation, model selection, ROC-AUC interpretation, gradient boosting, validation, and performance communication.

TOOLS

AutoMLXGBoostROC-AUCClassification
Visualization note: The visual above is a portfolio reconstruction based on the project methods and deliverables. It is provided to demonstrate the analysis without reproducing restricted course or organizational materials.