Publications by authors named "Alexandre Perez-Lebel"
Article Synopsis
- Large-scale databases, particularly in healthcare, often contain missing values that complicate analyses; however, these databases are valuable for training machine learning models aimed at tasks like forecasting and identifying biomarkers.
- A systematic benchmark of missing-value strategies was conducted using various health datasets, comparing native handling of missing values versus imputation methods, revealing that incorporating indicators for imputed values is crucial for accurate predictions.
- The findings suggest that leveraging machine learning methods that directly support missing values leads to better predictive outcomes and lower computational costs compared to traditional imputation techniques.
