Dependencies and Extras#
Core Install#
python -m pip install --upgrade pip
python -m pip install design-research-analysis
Editable contributor setup:
git clone https://github.com/cmudrc/design-research-analysis.git
cd design-research-analysis
python -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -e ".[dev]"
Or use:
make dev
Maintainer workflows target Python 3.12 from .python-version.
Extras Matrix#
Extra |
Purpose |
|---|---|
|
DataFrame/file profiling, schema validation, and codebook workflows |
|
Sequence and HMM workflows |
|
Default sentence-transformer backend for text embedding |
|
Language/topic modeling workflows |
|
Non-PCA projection backends and clustering for embedding-map workflows |
|
Legacy alias for |
|
Inferential and model-based statistics |
|
Convenience bundle for all analysis extras |
|
Contributor tooling |
Unified-table coercion, validation, and derived-column helpers are part of the
base install, so there is no separate table extra to add.
Base installs already support first-order Markov analysis, custom-embedder
language convergence, NumPy regression, PCA maps, and plotting. seq adds
HMM and graph backends. lang and embeddings add topic modeling and the
default sentence-transformer backend. maps adds manifold backends and
clustering. stats adds SciPy/statsmodels workflows. data is required by
profile_dataframe, validate_dataframe, and generate_codebook.
Recommended install profiles:
HMM-focused studies:
python -m pip install "design-research-analysis[seq]"language + embedding studies:
python -m pip install "design-research-analysis[lang,embeddings]"text-driven embedding maps:
python -m pip install "design-research-analysis[maps,embeddings]"numeric-feature embedding maps:
python -m pip install "design-research-analysis[maps]"inference + dataset studies:
python -m pip install "design-research-analysis[stats,data]"broad analysis workstation setup:
python -m pip install "design-research-analysis[all]"
The run-embedding-maps CLI embeds text by default, so its default path
needs [maps,embeddings]. With --feature-columns it skips text embedding,
and [maps] is enough. The dataset CLI commands profile-dataset,
validate-dataset, and generate-codebook require [data].
If you are working from a local checkout instead of PyPI, replace
design-research-analysis with . and add -e to install the same
extras in editable mode.
Release packaging validation is exposed via make release-check.