This repo has two git-backed components, slides/ and workshop/ . If you update the R dependencies in either, make sure to re-run rsconnect::writeManifest() in the respective directory.
Rendered output here: https://ccb.connect.hms.harvard.edu/missing_data/
Rendered workshop here: https://ccb.connect.hms.harvard.edu/missing_data_workshop/
Missingness in data is a ubiquitous and important issue in biomedical research, whether it’s an assay that fails, a subject that declines to answer a survey question, or a lost sample. When it comes time to analyze the data, the choices on handling these missing values can impact the downstream scientific conclusions.
This workshop will provide an introduction to important concepts, strategies, and tools in missing data analysis. We will also discuss several real-world examples of missing data analysis in biomedical research, including examples from epidemiological surveys, genomics, and single-cell multi-omics data. The first hour will be a seminar, the second will be a hands-on workshop where attendees run code.
How to:
- Assess the character of missingness in data
- Assess feasible modeling strategies for missing data
- Visualize missing data with
ggmice - Impute and model missing data with
miceandbrms
HMS graduate students, postdocs, or faculty who are interested in analyzing data with missing values.
The hands-on workshop requires an installation of R and several packages which can be installed with the following commands:
pkgs <- c("ggplot2", "dplyr", "mice", "ggmice", "brms")
install.packages(pkgs, Ncpus = 4)
Contact [email protected] with installation questions.