NEWS.md

In titaniumtroop/aRlegislation: A Dataset of Arkansas Legislation from 2001-2019

v0.2.0:

• Stored LDA, lawmaker, and lawmaker-session models individually in order to meet Github maximum size requirements
• Created function to load LDA models (calculations take forever, so this isn't something that will be generated frequently)
• Created functionss to load derivative models based on dot product of LDA models and sparse lawmaker matrices (both on a career and by-session basis)
• Excluded most files in these directories from builds to keep package size down (only topics 200-203 are included -- it was easiest to exclude 100-199); perhaps a download function should be written
• Created demo folder for machine learning analyses
• Excluded from Github sync for now pending possible publication of masters thesis
• Currently these are Rmd files that are used to run the analyses and generate images for use in thesis/presentations
• These could stand to be annotated/documented as standalone vignettes
• data-raw/results folder houses results of analyses; excluded from git due to large size of some of the files
• In order to run corruption_NN analyses, SVM analyses must be run first (the search grid is duplicated so some of the parameters carry over and can be compared)
• These analyses will be saved to the data-raw/results directory, which may need to be created

v0.1.8:

• Add corruption flag to lawmakers dataset (defaults to FALSE; set to TRUE only with comment showing link to news source)
• Considering whether to deprecate using sponsor short name as assigned during each session, and instead use a legislator_id based on the entity-resolved lawmakers

v0.1.7:

• Bug fixes for sponsor names
• Create nested dataset for entity-resolved lawmakers that nests their details for each session

v0.1.6:

• Manually updated missing sponsor full names and some other missing details

v0.1.5:

• Added legislation_corpus dataset
• Change seniority to match listed number (was multiplied by -1 so #1 seniority was highest; however, not using that field for anything currently, and shouldn't be confusing to future users)
• Further cleaning of sponsor details

v0.1.4:

• Moved README examples to vignettes

v0.1.3:

• Fix munging error that left footers in some act text

v0.1.2:

• Run all CRAN build checks to confirm that package works across multiple systems

v0.1.1:

• change column names in dataset to be tidy-compliant
• first README draft added

v0.1.0:

• Initial commit of data package that passes R CMD build checks

titaniumtroop/aRlegislation documentation built on May 4, 2020, 3:24 a.m.