- R coding example is added (sample.R) based on multiple received requests (2024-06-16)
- cMCA is from: Fujiwara and Liu, "Contrastive multiple correspondence analysis (cMCA): using contrastive learning to identify latent subgroups in political parties." PLOS ONE, vol. 18, no.7, e0287180 (20 pages), 2023. PLOS ONE, arXiv
- cMCA finds principal components along which a target categorical dataset has more variance when compared to a background categorical dataset.
- This implementation also integrates automatic selection of a contrast parameter (a hyperparameter of cMCA), which is originally developed for contrastive principal component analysis (cPCA) in: Fujiwara et al., "Network Comparison with Interpretable Contrastive Network Representation Learning", Journal of Data Science, Statistics, and Visualisation, 2022. JDSSV, arXiv
- Other use cases (using machine maintenance data) can be found in: Zhan et al., "A Visual Analytics Approach for the Diagnosis of Heterogeneous and Multidimensional Machine Maintenance Data." Proc. PacificVis, 2021. IEEE Xplore
- Python3
- R users can still utilize cMCA library by using reticulate package. Examples can be found in sample.R
- Note: Tested on macOS Sonoma, Ubuntu 22.0.4 LTS, and Windows 10.
-
Install with pip3. Move to the directory of this repository. Then,
pip3 install .
-
Import installed modules from python (e.g.,
from cmca import CMCA). See sample.py for examples. -
For detailed documentations, please see doc/index.html or directly see comments in cmca/cmca.py.
-
To use cMCA from R, you can utilize reticulate package. See sample.R
- Data type conversions are important for several case. For example:
- When identifying optional parameters of cMCA with R integer values, convert like
n_components=as.integer(2)(otherwise, 2 is considered as an index data type) - cMCA returns outputs as numpy objects in general. So, to convert in R native objects, use
py_to_r. Example code to convert numpy array to r matrix object can be found in sample.R (nparr_to_rmatfunction).
- When identifying optional parameters of cMCA with R integer values, convert like
- sample.R covers only a part of sample.py. Other remaining examples can be implemented if requested.
- Data type conversions are important for several case. For example:
Please, cite:
- Fujiwara and Liu, "Contrastive multiple correspondence analysis (cMCA): using contrastive learning to identify latent subgroups in political parties." PLOS ONE, vol. 18, no.7, e0287180 (20 pages), 2023.