We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.AP

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Statistics > Applications

Title: semopy 2: A Structural Equation Modeling Package with Random Effects in Python

Abstract: Structural Equation Modeling (SEM) is an umbrella term that includes numerous multivariate statistical techniques that are employed throughout a plethora of research areas, ranging from social to natural sciences. Until recently, SEM software was either commercial or restricted to niche languages, and the lack of SEM packages compatible with more mainstream programming languages was dire. To combat that, we introduced a Python package semopy 1 that surpassed other state-of-the-art software in terms of performance and estimation accuracy. Yet, it was lacking in functionality and its usage was burdened with unnecessary boilerplate code. Here, we introduce a complete overhaul of semopy that improves upon the previous results and comes with lots of new capabilities. Furthermore, we propose a novel SEM model that combines in itself a notion of random effects from linear mixed models (LMMs) to model numerous phenomena, such as spatial data, time series or population stratification in genetics.
Subjects: Applications (stat.AP); Mathematical Software (cs.MS)
Cite as: arXiv:2106.01140 [stat.AP]
  (or arXiv:2106.01140v3 [stat.AP] for this version)

Submission history

From: Georgy Meshcheryakov [view email]
[v1] Wed, 2 Jun 2021 13:18:03 GMT (237kb,D)
[v2] Thu, 3 Jun 2021 10:20:12 GMT (237kb,D)
[v3] Wed, 9 Jun 2021 17:35:07 GMT (245kb,D)

Link back to: arXiv, form interface, contact.