What question did this study set out to answer?

To evaluate how different data balancing techniques influence classification performance and determine whether class imbalance alone or co-occurring data complexities hinder learning algorithms.

June 1, 2004ACM SIGKDD Explorations Newsletter

A study of the behavior of several methods for balancing machine learning training data

View Full Paper

Ask AI

Bookmark

Authors

GBGustavo E. A. P. A. BatistaUNSW Sydney RPRonaldo C. PratiUniversidade Federal do ABC MMMaria Carolina MonardUniversidade de São Paulo

Discussion

Member takes

Implication

Comparative evaluation demonstrates over-sampling outperforms under-sampling across imbalanced benchmark datasets, suggesting minority sample sparsity and class overlap primarily drive learning...

Key Points

To evaluate how different data balancing techniques influence classification performance and determine whether class imbalance alone or co-occurring data complexities hinder learning algorithms.
Evaluated ten balancing algorithms, including standard under-sampling, random over-sampling, and combinations of SMOTE with data cleaning techniques (SMOTE + Tomek and SMOTE + ENN).
Tested methods across thirteen benchmark datasets from the UCI Machine Learning Repository.
Assessed model performance using the area under the receiver operating characteristic curve (AUC) and quantified the syntactic complexity of the resulting decision trees.
Over-sampling approaches generally yielded superior AUC performance compared to under-sampling methods across the evaluated datasets.
Methods pairing SMOTE with Tomek links or edited nearest neighbors (ENN) achieved strong performance on datasets with few minority instances by reducing class overlap.
Over-sampled training data increased decision tree complexity relative to original data, with random over-sampling showing the smallest increase in rule count and SMOTE + ENN the smallest increase in conditions per rule.

Cite This Study

Batista et al. (2004) studied this question.

synapsesocial.com/papers/69d7fbfc66a29169b4bedb38 https://doi.org/10.1145/1007730.1007735

View Full Paper

Ask AI

Bookmark

June 1, 2004ACM SIGKDD Explorations Newsletter

A study of the behavior of several methods for balancing machine learning training data

View Full Paper

Ask AI

Bookmark

Authors

GBGustavo E. A. P. A. BatistaUNSW Sydney RPRonaldo C. PratiUniversidade Federal do ABC MMMaria Carolina MonardUniversidade de São Paulo

Discussion

Member takes

Implication

Comparative evaluation demonstrates over-sampling outperforms under-sampling across imbalanced benchmark datasets, suggesting minority sample sparsity and class overlap primarily drive learning...

Key Points

To evaluate how different data balancing techniques influence classification performance and determine whether class imbalance alone or co-occurring data complexities hinder learning algorithms.
Evaluated ten balancing algorithms, including standard under-sampling, random over-sampling, and combinations of SMOTE with data cleaning techniques (SMOTE + Tomek and SMOTE + ENN).
Tested methods across thirteen benchmark datasets from the UCI Machine Learning Repository.
Assessed model performance using the area under the receiver operating characteristic curve (AUC) and quantified the syntactic complexity of the resulting decision trees.
Over-sampling approaches generally yielded superior AUC performance compared to under-sampling methods across the evaluated datasets.
Methods pairing SMOTE with Tomek links or edited nearest neighbors (ENN) achieved strong performance on datasets with few minority instances by reducing class overlap.
Over-sampled training data increased decision tree complexity relative to original data, with random over-sampling showing the smallest increase in rule count and SMOTE + ENN the smallest increase in conditions per rule.

Cite This Study

Batista et al. (2004) studied this question.

synapsesocial.com/papers/69d7fbfc66a29169b4bedb38 https://doi.org/10.1145/1007730.1007735

Also Consider

Synapse has enriched 2 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark