The quollr package provide functions to construct a model in the 2D space based on 2D embedding and then lifts to the high dimensional space. The package also provides a visualization that integrates this model with high dimensional data using the tour technique. For a thorough background and discussion on this work, please read our paper [link to the paper].
Installation
You can install the released version of quollr from CRAN with:
# install.packages("quollr")And the development version from GitHub with:
# install.packages("remotes")
# remotes::install_github("JayaniLakshika/quollr")Usage
Our approach involves dividing the high dimensional data set into two parts: a training set to construct the model and a test set to generate predictive values and residuals. To implement our approach, first we use a 2D embedding data set as the initial point. The output of our algorithm is a tour that displays the model and original high dimensional data in the high-dimensional space. Our algorithm comprises two main phases:(1) generate the model in the 2D space, and (2) generate the model in the high-D space.
Roadmap
- To learn more about the basic steps of the algorithm: 1. Main steps
- To create the full hexagonal grid: 2. Visualise full hexagonal grid
- To generate 2D embeddings for test data: 3. Predict 2D embeddings
Copyright
This package is licensed under the MIT license.