Supplementary material - Assessment of the accuracy of machine learning models to classify topic and sentiment of Canadian tweets about public health measures during COVID-19 using Bayesian latent class analysis

The objective of this study is to estimate the sensitivity and specificity of machine learning models to correctly identify the topics and sentiment of Canadian posts on Twitter, between December 2020 and February 2021 or between June and August 2021, related to COVID-19 public health measures, using human annotation as imperfect additional tests in Bayesian latent class analysis models. This dataset included anonymized data, cross tabulation of the tests' results, and codes used for the analyses.