structural-topic-model 0.2.2__tar.gz → 0.3.0__tar.gz

{structural_topic_model-0.2.2 → structural_topic_model-0.3.0}/.gitignore

@@ -33,7 +33,6 @@ htmlcov/
 .vscode/
 
 # Project-specific
-stm*
-!pystm/stm.py
+stm_r*
 application/
 uv.lock

{structural_topic_model-0.2.2 → structural_topic_model-0.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: structural-topic-model
-Version: 0.2.2
+Version: 0.3.0
 Summary: Python implementation of the Structural Topic Model (STM), a port of the R stm package with a scikit-learn style API
 Project-URL: Homepage, https://github.com/hirata-keisuke/pystm
 Project-URL: Repository, https://github.com/hirata-keisuke/pystm
@@ -22,7 +22,7 @@ Requires-Dist: scikit-learn>=1.9.0
 Requires-Dist: scipy>=1.14
 Description-Content-Type: text/markdown
 
-# pystm — Structural Topic Model in Python
+# stm — Structural Topic Model in Python
 
 [![PyPI](https://img.shields.io/pypi/v/structural-topic-model)](https://pypi.org/project/structural-topic-model/)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
@@ -45,7 +45,7 @@ pip install structural-topic-model
 
 ```python
 import numpy as np
-from pystm import StructuralTopicModel
+from stm import StructuralTopicModel
 
 # X: (n_docs, n_vocab) word count matrix (dense or scipy.sparse)
 # covar: (n_docs, n_covariates) prevalence covariate matrix (intercept added automatically)
@@ -94,7 +94,7 @@ Estimated via Distributed Poisson regression (equivalent to the R package's defa
 Regress topic proportions on covariates using method of composition, returning coefficients with measurement uncertainty:
 
 ```python
-from pystm import estimate_effect
+from stm import estimate_effect
 
 eff = estimate_effect(model, covar, uncertainty="Global", nsims=25)
 tables = eff.summary()      # {topic: structured array with estimate/std_error/t_value/p_value}
@@ -104,7 +104,7 @@ tables[0]["estimate"]       # coefficients for topic 0 (first entry is intercept
 ## Topic Selection (searchK)
 
 ```python
-from pystm import search_k
+from stm import search_k
 
 res = search_k(X, K_values=[5, 10, 15], prevalence=covar,
                model_params={"max_iter": 100})
@@ -117,7 +117,7 @@ res["exclus"]    # exclusivity
 ## Diagnostics
 
 ```python
-from pystm import topic_corr, semantic_coherence, exclusivity, check_residuals
+from stm import topic_corr, semantic_coherence, exclusivity, check_residuals
 
 tc = topic_corr(model, cutoff=0.01)    # topic correlation graph (simple method)
 tc.posadj                               # positive correlation adjacency matrix

{structural_topic_model-0.2.2 → structural_topic_model-0.3.0}/README.md

@@ -1,4 +1,4 @@
-# pystm — Structural Topic Model in Python
+# stm — Structural Topic Model in Python
 
 [![PyPI](https://img.shields.io/pypi/v/structural-topic-model)](https://pypi.org/project/structural-topic-model/)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
@@ -21,7 +21,7 @@ pip install structural-topic-model
 
 ```python
 import numpy as np
-from pystm import StructuralTopicModel
+from stm import StructuralTopicModel
 
 # X: (n_docs, n_vocab) word count matrix (dense or scipy.sparse)
 # covar: (n_docs, n_covariates) prevalence covariate matrix (intercept added automatically)
@@ -70,7 +70,7 @@ Estimated via Distributed Poisson regression (equivalent to the R package's defa
 Regress topic proportions on covariates using method of composition, returning coefficients with measurement uncertainty:
 
 ```python
-from pystm import estimate_effect
+from stm import estimate_effect
 
 eff = estimate_effect(model, covar, uncertainty="Global", nsims=25)
 tables = eff.summary()      # {topic: structured array with estimate/std_error/t_value/p_value}
@@ -80,7 +80,7 @@ tables[0]["estimate"]       # coefficients for topic 0 (first entry is intercept
 ## Topic Selection (searchK)
 
 ```python
-from pystm import search_k
+from stm import search_k
 
 res = search_k(X, K_values=[5, 10, 15], prevalence=covar,
                model_params={"max_iter": 100})
@@ -93,7 +93,7 @@ res["exclus"]    # exclusivity
 ## Diagnostics
 
 ```python
-from pystm import topic_corr, semantic_coherence, exclusivity, check_residuals
+from stm import topic_corr, semantic_coherence, exclusivity, check_residuals
 
 tc = topic_corr(model, cutoff=0.01)    # topic correlation graph (simple method)
 tc.posadj                               # positive correlation adjacency matrix

{structural_topic_model-0.2.2 → structural_topic_model-0.3.0}/README_ja.md

@@ -1,4 +1,4 @@
-# pystm — Python による構造的トピックモデル
+# stm — Python による構造的トピックモデル
 
 [![PyPI](https://img.shields.io/pypi/v/structural-topic-model)](https://pypi.org/project/structural-topic-model/)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
@@ -21,7 +21,7 @@ pip install structural-topic-model
 
 ```python
 import numpy as np
-from pystm import StructuralTopicModel
+from stm import StructuralTopicModel
 
 # X: (n_docs, n_vocab) の単語カウント行列（dense / scipy.sparse どちらも可）
 # covar: (n_docs, n_covariates) の prevalence 共変量（切片は自動付与）
@@ -70,7 +70,7 @@ R 版の `kappa.prior="L1"`（既定）に相当する Distributed Poisson 回
 トピック比率を目的変数とする回帰を method of composition（変分事後分布からの θ サンプリング）で行い、測定不確実性込みの係数を返します：
 
 ```python
-from pystm import estimate_effect
+from stm import estimate_effect
 
 eff = estimate_effect(model, covar, uncertainty="Global", nsims=25)
 tables = eff.summary()      # {topic: 構造化配列（estimate/std_error/t_value/p_value）}
@@ -80,7 +80,7 @@ tables[0]["estimate"]       # トピック 0 の回帰係数（先頭が切片
 ## トピック数の選択（searchK 相当）
 
 ```python
-from pystm import search_k
+from stm import search_k
 
 res = search_k(X, K_values=[5, 10, 15], prevalence=covar,
                model_params={"max_iter": 100})
@@ -93,7 +93,7 @@ res["exclus"]    # 排他性
 ## 診断
 
 ```python
-from pystm import topic_corr, semantic_coherence, exclusivity, check_residuals
+from stm import topic_corr, semantic_coherence, exclusivity, check_residuals
 
 tc = topic_corr(model, cutoff=0.01)    # トピック相関グラフ（simple 法）
 tc.posadj                               # 正相関の隣接行列

{structural_topic_model-0.2.2 → structural_topic_model-0.3.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "structural-topic-model"
-version = "0.2.2"
+version = "0.3.0"
 description = "Python implementation of the Structural Topic Model (STM), a port of the R stm package with a scikit-learn style API"
 readme = "README.md"
 requires-python = ">=3.12"
@@ -46,8 +46,11 @@ dev = [
 requires = ["hatchling"]
 build-backend = "hatchling.build"
 
+[tool.pytest.ini_options]
+pythonpath = ["."]
+
 [tool.hatch.build.targets.wheel]
-packages = ["pystm"]
+packages = ["stm"]
 
 [tool.hatch.build.targets.sdist]
-only-include = ["pystm", "tests", "scripts", "README.md", "README_ja.md", "LICENSE"]
+only-include = ["stm", "tests", "scripts", "README.md", "README_ja.md", "LICENSE"]

{structural_topic_model-0.2.2 → structural_topic_model-0.3.0}/scripts/validate_gadarian.py

@@ -38,10 +38,10 @@ sys.path.insert(0, str(Path(__file__).parent))
 from gadarian_prep import load_gadarian, prep_documents, text_processor
 
 sys.path.insert(0, str(Path(__file__).parent.parent))
-from pystm import StructuralTopicModel, estimate_effect
-from pystm._estep import estep
-from pystm._mstep import opt_beta, opt_mu, opt_sigma
-from pystm._utils import row_softmax, to_doc_list
+from stm import StructuralTopicModel, estimate_effect
+from stm._estep import estep
+from stm._mstep import opt_beta, opt_mu, opt_sigma
+from stm._utils import row_softmax, to_doc_list
 
 PASS = []
 

{structural_topic_model-0.2.2 → structural_topic_model-0.3.0}/tests/test_features.py

@@ -3,7 +3,7 @@
 import numpy as np
 import pytest
 
-from pystm import (
+from stm import (
     StructuralTopicModel,
     check_residuals,
     estimate_effect,

{structural_topic_model-0.2.2 → structural_topic_model-0.3.0}/tests/test_stm.py

@@ -1,4 +1,4 @@
-"""Tests for pystm.StructuralTopicModel.
+"""Tests for stm.StructuralTopicModel.
 
 Validation strategy (no R available for direct comparison):
 - ELBO increases over EM iterations and converges,
@@ -11,7 +11,7 @@ import numpy as np
 import pytest
 from scipy.optimize import linear_sum_assignment
 
-from pystm import StructuralTopicModel
+from stm import StructuralTopicModel
 
 
 def make_corpus(n_docs=200, n_topics=3, vocab_size=60, doc_len=80,