Have you ever heard of missing data? Do you ever crave the magic of PCA? If you do, you've found your Home. Here, we bring you a few methods to deal with missing data effectively so that you can run your PCA smoothly!
This method was proposed by Troyanskaya et al.. In essence, it runs iteratively runs regression imputation using the top PCs obtained until convergence. Another imputation method is used as initialization.
This is a more conventional method that is also mentioned by Troyanskaya et al.. It imputes by finding a weighted average from its nearest neighbors. This method does not require PCA: it can be a general-purpose imputation method.
This is a newly developed method proposed by Podani et al.. It skips the problem of imputation entirely: namely, it attempts to perform PCA without imputing. It uses the pairwise complete observation to compute Eigen-decomposition and then set all missing values to 0 while computing the scores.
You won't need other packages. All you need to do is to source the scripts and run the methods.
Podani, J., Kalapos, T., Barta, B., & Schmera, D. (2021). Principal component analysis of incomplete data-A simple solution to an old problem. Ecological Informatics, 61, 101235.
Troyanskaya, O., Cantor, M., Sherlock, G., Brown, P., Hastie, T., Tibshirani, R., ... & Altman, R. B. (2001). Missing value estimation methods for DNA microarrays. Bioinformatics, 17(6), 520-525.