We are delighted to announce the recipients of the 2024 Award for Innovation in Statistical Programming and Analytics (AISPAA). This award recognizes individuals and teams who have made significant contributions to the development and enhancement of tools, software, and methodologies that address problems in statistical programming and analytics. It also honors those who have made large contributions to the community of statistical programmers and analysts.
This year, we proudly present the award to two outstanding recipients:
Dr. Byron C. Jaeger
Department of Biostatistics and Data Science, Wake Forest University School of Medicine
Dr. Jaeger is honored for his pioneering work in developing methods for oblique random forests, which he has effectively disseminated through his R package, aorsf. This package is the first to support oblique random forests for classification, regression, and survival analysis. It also introduces novel methods for variable importance and selection. Dr. Jaeger's 'accelerated' models enable aorsf to perform as quickly as other popular R packages such as ranger and randomForestSRC. He has optimized the code in aorsf for efficient computation of partial dependence and individual conditional expectation curves. Since its release, aorsf has been downloaded 14,000 times from CRAN and has received 5 citations for its companion paper since 2023.
Dr. Cynthia Rudin, Yingfan Wang, and Haiyang Huang
Department of Computer Science, Electrical and Computer Engineering, Statistical Science, Mathematics, and Biostatistics & Bioinformatics - Duke University
Dr. Rudin and her team are recognized for their development of PaCMAP (Pairwise Controlled Manifold Approximation), a dimensionality reduction method designed for visualization that preserves both local and global structures of data in its original space. PaCMAP features a simpler loss function than previous approaches, making it more accessible and easier to troubleshoot. It is robust to hyperparameter selection and initial conditions due to its loss terms that maintain global structure and stable cluster locations. The method's versatility is evident as it is available in both Python and R, with fewer dependencies than most other dimensionality reduction methods. PaCMAP also offers interoperability by sharing an API with existing machine learning pipelines like scikit-learn. Its efficiency is demonstrated by its faster performance, scaling to large datasets through the use of "mid-near pairs." PaCMAP has gained significant recognition, boasting over 460 GitHub stars and an active user base. It has been applied in various fields, including the analysis of Parkinson's disease, name-ethnicity classification, and natural language embeddings.
We congratulate Dr. Jaeger and Dr. Rudin's team on their remarkable achievements and their invaluable contributions to the field of statistical programming and analytics. Their innovative work exemplifies the spirit of the AISPAA, and we are excited to see the continued impact of their contributions on the community.
Please join us in celebrating the accomplishments of this year’s AISPAA recipients!