Computational Statistics and Data Analytics Lab

CSDA (Computational Statistics and Data Analytics) is a multidisciplinary student-centered research group, working on research projects in applied statistics, machine learning, deep learning, and AI. The following are the main areas of research (but not limited to):

• Statistical monitoring: statistical methods for system monitoring purposes, including control charts, health monitoring, profile monitoring, IoT monitoring, image monitoring, etc.
• Statistics for AI: AI model compression techniques and optimization: fine-tuning, knowledge distillation, pruning, quantization.
• Uncertainty Quantification: conformal predictors and robust statistics
• Health Analytics: sports medicine/orthopedics, cancer flow cytometry, cancer x metabolic syndromes, predictive modeling and causal inference.
• Education Analytics: AI advising chatbot, RAG systems, LLM, Agents.

CSDA Lab Meetings in Spring 2025

Methods and Technologies

The CSDA Lab is primarily a data analytics group that uses a wide range of statistical and mathematical methods, including:

• Exploratory Statistical Analysis: Principal component analysis (PCA), multiple correspondence analysis (MCA), and more.
• Statistical Modeling: Poisson regression, Negative Binomial regression, Zero-Inflated models, Gamma regression, Mixed models, Time Series, hypothesis testing, Cox models, and more.
• Clustering techniques: k-means, latent class analysis (LCA), and more.
• Classification: SVM, regressions, Random forest, and more.
• Predictive models: RNN, CNN, regression, LSTM, and more.
• Imputation methods: MICE, EM, and more.
• Signal processing: Wavelet analysis, Fourier transforms, multiscale analysis, and more.
• Tools: R, Quarto, Python, SAS, and MATLAB.

Grants and Collaboration

• IHMC and ARL
• Air Force Research Laboratory
• Mayo Clinic
• Andrews Research & Education Foundation

Research Guidelines

Please review the research guidelines.

Joining CSDA Lab

If you are interested in joining, please go to the recruitment page.