Posts
Gaussian process regression as an alternative to continuous-time autoregression
For small datasets, Gaussian process regression can be just as effective.
Estimating correlation matrices when the data include missing and left-censored values
A one-step Bayesian approach.
Probabilistic principal component analysis for censored data
Application of Bayesian PCA to a heavily left-censored (simulated) dataset.
A Bayesian approach to predicting disinfection byproduct formation using regularizing priors
Using heirarchical shrinkage priors on linear regression coefficients to reduce the variance of ccorrelated predictors.
Regression models with censored predictors
Another modification of brms-generated Stan code, this time to fit regression models with censored predictors.
Building a continuous-time autoregressive model in brms
A simple modification of brms-generated Stan code to fit first-order autoregressive models to irregularly-spaced time series.
Diagnostics for censored autoregressive models fitted with brms
Posterior predictive checks and simulated residuals.
A brownification pause?
Censored autoregression with a smoothed time covariate.
New preprint!
Aluminum in drinking water can interact with orthophosphate, increasing lead solubility.
Comparing water quality time series using a generalized additive mixed model
Revisiting work from 2016 to better model time series with non-linear trends.
Non-parametric matched pair testing with left-censored data
Comparing two groups of measurements when some values are below one or multiple detection limit(s).
Peak detection for qualitative XRD analysis in R
An improved workflow for visualizing X-ray diffraction data in R.
Lead solubility prediction using R and Shiny
An R package and Shiny app for PHREEQC-based equilibrium lead and copper solubility prediction.