HyperTools: a Python Toolbox for Gaining Geometric Insights into High-Dimensional Data

Authors: Andrew C. Heusser, Kirsten Ziman, Lucy L. W. Owen, Jeremy R. Manning

JMLR 2017 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We highlight several examples of how low-dimensional projections may be used to understand the geometry of high-dimensional data in Figure 1. [...] The Hyper Tools toolbox (current version: 0.4.2) provides a powerful set of Python functions for projecting high dimensional data onto lower-dimensional spaces, aligning data of different types, and visualizing the results in publication-quality figures and movies.
Researcher Affiliation Academia Andrew C. Heusser EMAIL Kirsten Ziman EMAIL Lucy L. W. Owen EMAIL Jeremy R. Manning EMAIL Department of Psychological and Brain Sciences, Dartmouth College, Hanover, NH 03755, USA
Pseudocode No The paper provides examples of Python function calls (e.g., hyp.plot(...)) to demonstrate the toolbox's usage, but it does not contain structured pseudocode or algorithm blocks describing the underlying methods.
Open Source Code Yes Hyper Tools is open-source, installable from Git Hub or pip (pip install hypertools), and is distributed with the MIT License.
Open Datasets Yes Figure 1: Data visualization examples. [...] b. Topic modeling (Blei et al., 2003) of political Twitter data: ... (link to data). c. Changing temperatures across the Earth s surface from 1875 2013 (link to data). d. Brain/movie trajectories during movie viewing (link to data).
Dataset Splits Yes (Top left) Group-averaged trajectories of brain activity from the ventral visual cortex, split into two randomly-selected groups of subjects (group 1: n = 6, group 2: n = 5) watching Raiders of the Lost Ark (Haxby et al., 2011).
Hardware Specification No The paper describes a software toolbox and its functionalities but does not provide any specific details about the hardware (e.g., CPU, GPU models, memory) used for running experiments or generating the results shown.
Software Dependencies No The toolbox depends on the following open-source software packages: Matplotlib (Hunter, 2007) for plotting functionality, Seaborn (Waskom et al., 2016) for plot styling, Scikit-learn (Pedregosa et al., 2011) for data analysis (dimensionality reduction, clustering, etc.), and PPCA for inferring missing data (Tipping and Bishop, 1999).
Experiment Setup Yes hyp.plot(list of arrays, reduce= TSNE , align= hyper , ndims=3) This example function call projects the high-dimensional data onto 3 dimensions using the t-SNE algorithm, aligns the data matrices in the given list of arrays into a common space using hyperalignment, and produces a 3D plot analogous to those shown in Figure 1.