This repo contains Shiny app code which is a proof of concept for a data validation report generated from the data.validator package. It is also a test run of our shiny.fluent package, which allows for building Shiny apps with React.js components without the need for JS programming.
You can view the application here: https://pawelp.shinyapps.io/data-validator-report/
In our projects, we perform rigorous data validation before we use data, especially when the data is updated periodically or can be uploaded by the user. Data almost always contains some errors, no matter the source. Most of these errors are related to the format, but often there are also quantitative requirements. But that's not all - additionally, there are very important expert rules which are often project specific. You can read about these rules in our blogpost about data quality and how we saved significant money for our clients by adhering to them.
An important part of our validation is close collaboration with the client and project stakeholders. There are often situations where the validation rule depends on expert knowledge or a specific business decision. To make this collaboration efficient, we present validation results as a report and share them with the client. Thanks to the report, it is easier to inspect corrupted data and discuss potential root causes and solutions.
I used shiny.fluent package to build a proof of concept that:
- proves how easy it can be to use React.js components within Shiny
- has a professional-looking UI thanks to Microsoft Fluent UI components
Note: this POC app is a hypothetical report generated after running data_validation.R code.
Data.validator is similar to testthat, but designed for data validation. It is built on top of the awesome assertr package, with a simplified API for generating reports. The most common use case is to combine data.validator code with data processing scripts used in ETLs / pipelines / batch jobs.
See the repo for more examples.
Note: Current version which is released on CRAN (0.1.5) generates rmarkdown HTML report using shiny.semantic. This POC shiny.fluent app is not integrated yet.
At first glance it might seem like any other package with regular HTML components themed with CSS and JS. The difference is that shiny.fluent uses React.js components, which is a different approach. As a first library with React.js components ported to Shiny we drew on Microsoft Fluent UI.
See the repo for example apps built with shiny.fluent.