Constellate User Guide

Data Tables

If a Python cell outputs a Pandas DataFrame, then Constellate will show the table using a fairly sophisticated component. Some features to note:

• Numbers, booleans, datetimes, currencies, and JSON are automatically recognized: note how the numeric columns, including the dollar values, are right-aligned while everything else is left-aligned. This also allows for intelligent sorting behavior.
• Pagination is automatic, allowing for a relatively efficient render even for large datasets.
  • Note that, while Constellate's DataFrame rendering scales relatively well, it still keeps the entire dataset in memory and sends it to the client when they load the page. Anything not on the first page doesn't have to be rendered, but it does have to be sent to the browser, which will cripple load times for very large datasets. Instead, use a Tabulator widget as a Panel plot, which can send a server request to get more data.

Like Plotly, Constellate has to guess what the DataFrame value is: there's no "current DataFrame" feature in Pandas. The heuristic Constellate applies is simple: the last non-empty line is assumed to be the one that produces the DataFrame. This is usually correct, but is not bulletproof, and so be aware that you will most likely run into issues if you do things like display a DataFrame manually instead of simply having it be the last expression. To make sure Constellate knows exactly what to do, give it a directive: #constellate: dataframe [expr] where expr is a Python expression to be run after the cell that produces a DataFrame. The code for this Star would produce the same results if I had started with #constellate: dataframe or #constellate: dataframe exch.