Skip to content
This repository has been archived by the owner on Apr 11, 2024. It is now read-only.

Latest commit

 

History

History

Turkia

README.md

Modeling the Effects of Nutrition with Mixed-Effect Bayesian Network

This work proposes Mixed-Effect Bayesian Network (MEBN) as a method for modeling the effects of nutrition. It allows to identify both typical and personal correlations between nutrients and their bodily responses. Predicting a personal network of nutritional reactions would allow interesting applications at personal diets and in understanding this complex system. Brief theory of MEBN is given followed by implementation in R and Stan. A real life dataset from a nutritional Sysdimet study is then analyzed with the method and the results are visualized with an interactive JavaScript-visualization.

View HTML-version of the notebook here

Contents

Main body of the presentation is found here as fully functional RMarkdown notebook, and also HTML and PDF renderitions of it. The notebook uses my R-package "MEBN" for constructing a Mixed-Effect Bayesian Network that models the effects of nutrition from a real life nutritional dataset. The fully Bayesian estimation of the local mixed-effect models at the network is done with Stan. Visualization of the network is created with JavaScript library sigma.js and some customizations of it.

  • PersonalEffectsOfNutrition.Rmd : RMarkdown notebook
  • PersonalEffectsOfNutrition.html : HTML knit of the notebook with active JavaScript visualization
  • PersonalEffectsOfNutrition.pdf : PDF knit of the notebook
  • population_graph.htm : HTML document for graph visualization
  • biblio.bib : References for the notebook in BibLatex format
  • data-folder: The dataset that is used in the notebook analysis
  • Data description.csv : CSV file that describes the metadata for analyzed dataset
  • mebn-folder: The R-code for constructing MEBN and stan-model definitions
  • visualization-folder: JavaScript code for graph visualization
  • models-folder: Sampled Stan-models are cached in this folder once the notebook is executed. It is empty by default.

License

Material in this repository is licensed under CC 4.0 https://creativecommons.org/licenses/by/4.0/

Requirements

Besides working installation of Stan, following R-packages are required for this notebook to execute correctly.

Use install_packages.r script to install.

** Session Info **

R version 3.4.4 (2018-03-15) Platform: x86_64-w64-mingw32/x64 (64-bit) Running under: Windows Server 2008 R2 x64 (build 7601) Service Pack 1

Matrix products: default

locale: [1] LC_COLLATE=Finnish_Finland.1252 LC_CTYPE=Finnish_Finland.1252 [3] LC_MONETARY=Finnish_Finland.1252 LC_NUMERIC=C [5] LC_TIME=Finnish_Finland.1252

attached base packages: [1] stats graphics grDevices utils datasets methods base

loaded via a namespace (and not attached): [1] Rcpp_0.12.16 digest_0.6.15 rprojroot_1.3-2 plyr_1.8.4 grid_3.4.4 [6] gtable_0.2.0 backports_1.1.2 magrittr_1.5 evaluate_0.10.1 scales_0.5.0 [11] pillar_1.2.1 ggplot2_2.2.1 rlang_0.2.0 stringi_1.1.7 lazyeval_0.2.1 [16] rmarkdown_1.9 tools_3.4.4 stringr_1.3.0 munsell_0.4.3 yaml_2.1.19 [21] compiler_3.4.4 colorspace_1.3-2 htmltools_0.3.6 knitr_1.20 tibble_1.4.2