Machine learning project—cryptocurrency prediction model with feature engineering and evaluation
Developed and validated a cryptocurrency price prediction model using machine learning techniques. Performed walk-forward validation to evaluate model performance across multiple months of data and timeframes. Conducted feature engineering and data preparation workflows to improve predictive inputs. • Built a LightGBM model for price forecast resolution across multiple cryptocurrency timeframes • Used SQL to clean and prepare datasets • Engineered features by reducing 321k candidates to 2930 via multi-stage pruning • Visualized results using Power BI for interpretation and communication
2025 - 2026