Smarter numeric profiling​

Our numeric profiling histograms were created with simple defaults to get the application up and running and we wanted to make this visualization more useful. Our latest release adds several nice upgrades to understanding these data types.

• Meaningful data shapes — The number of buckets has been updated to better reflect the distribution of the data. Instead of displaying 40 buckets by default on all histograms, we are using the Freedman Diaconis Estimator. This algorithm is robust to capturing outliers and accounts for the size of the data set.

• Understanding the edges — A rug-plot below the histogram detail makes it is easy to pick out the standard deviation for outliers and identify unexpected clumps of data.

Start exploring data faster​

We don’t want the little things to get in the way of sharing the power of Rill’s data exploration with you. With this release we focused on making installing Rill faster and easier so you can get out of installer dialogues and back to exploring your data.

• Faster install times — We have used vercel/pkg to build binaries for different OSes and architectures. This will bundle the Rill Developer application as a single file, including a build of DuckDB, to make it faster than ever to get started. We currently support installing the binary through asset download links in our Github release change log.

• Clearer documentation — Our readme keeps growing and our docs need to evolve with our features. This week we have united our readme and developer-guide into markdown files that will become our source for documentation.