Heavy-duty time-series charts

Question

I have a bunch of custom log files where each line has a timestamp, and some data points. I actually began writing my own macOS program to import and visualize these logs, but the sheer number of data points makes any charting package I try basically throw up. Each log file has tens of thousands of lines/datapoints (sometimes even hundreds of thousands, actually).

So, what could anyone recommend to visualize this data? I'm looking for mostly line/scatter charts, and don't have an issue transforming the data into some other form, as I already have a parser written.

It would be a bonus if the generated charts were interactive, so I could zoom around them etc., rather than just a static image, for example.

Answer 1

ZingChart is a JavaScript charting library that supports big data sets, time-series formatting, and zooming/interactivity. You can see data comparisons here:

https://www.zingchart.com/demos/zingchart-vs/

Line charts and scatter plots are just a few of the chart types we offer. For information on how to convert Unix timestamps into readable dates, see our section on time-series formatting (scales, tooltips, value boxes, crosshairs) and our section on time-series scales.

As for chart interactivity, we support chart zooming, scrolling, and preview. (If you want the chart zoomed in to specified scale values at chart render, make sure to use the zoomToValues array in your scale object(s).) We also support interactive crosshairs, which you may be interested in if you're creating a lot of time-series line and scatter charts.

Disclosure: I'm a member of the ZingChart team, happy to answer any questions about our charting library.

Answer 2

Pandas + MatPlotLib to the rescue:

• Handles big data up to the size to require HDF5 storage
• Interactive Charts with zoom, pan, print, save, etc.
• Free, Gratis & Open Source
• Cross Platform

Some examples here.

For more interactivity you can also use bokeh which can also run in a Jupyter Notebook, examples & you might find it worth looking at HoloViews which lets you annotate your data and let it plot itself - there is a good example using several months worth of Earthquake data here.