[Paper Review] Plotly-Resampler: Effective Visual Analytics for Large Time Series

Jonas Van Der Donckt, Jeroen Van Der Donckt+2 more|arXiv (Cornell University)|Jun 17, 2022

Data Visualization and Analytics4 citations

TL;DR

Plotly-Resampler is an open-source Python library that enables interactive, scalable visualization of large time series by dynamically aggregating data based on the current view in a Plotly dashboard. It achieves sub-second responsiveness even with millions of data points by offloading aggregation to a back-end, significantly outperforming alternatives in scalability and interactivity benchmarks.

ABSTRACT

Visual analytics is arguably the most important step in getting acquainted with your data. This is especially the case for time series, as this data type is hard to describe and cannot be fully understood when using for example summary statistics. To realize effective time series visualization, four requirements have to be met; a tool should be (1) interactive, (2) scalable to millions of data points, (3) integrable in conventional data science environments, and (4) highly configurable. We observe that open source Python visualization toolkits empower data scientists in most visual analytics tasks, but lack the combination of scalability and interactivity to realize effective time series visualization. As a means to facilitate these requirements, we created Plotly-Resampler, an open source Python library. Plotly-Resampler is an add-on for Plotly's Python bindings, enhancing line chart scalability on top of an interactive toolkit by aggregating the underlying data depending on the current graph view. Plotly-Resampler is built to be snappy, as the reactivity of a tool qualitatively affects how analysts visually explore and analyze data. A benchmark task highlights how our toolkit scales better than alternatives in terms of number of samples and time series. Additionally, Plotly-Resampler's flexible data aggregation functionality paves the path towards researching novel aggregation techniques. Plotly-Resampler's integrability, together with its configurability, convenience, and high scalability, allows to effectively analyze high-frequency data in your day-to-day Python environment.

Motivation & Objective

• Address the lack of scalable, interactive time series visualization tools in Python data science workflows.
• Overcome the trade-off between interactivity and performance when rendering millions of data points.
• Enable seamless integration with existing Python environments like Jupyter notebooks and Dash applications.
• Provide a highly configurable, extensible platform for visual analytics and research into time series aggregation techniques.
• Ensure responsive, real-time interaction during zooming and panning through efficient client-server data updates.

Proposed method

• Integrates as a decorator (FigureResampler) on top of Plotly's Python visualization backend.
• Splits visualization into a front-end (displaying aggregated data) and back-end (storing full dataset and processing view changes).
• Uses dynamic callbacks: on zoom/pan events, the front-end sends the new view range (relayoutData) to the back-end.
• The back-end performs on-demand data aggregation using an efficient heuristic of the LTTB algorithm by default.
• Sends only the updated, aggregated data (updateData) back to the front-end for rendering, minimizing network overhead.
• Indicates aggregated data via a [R] prefix and estimated bin size in the legend to prevent misinterpretation of downsampled data.

Experimental results

Research questions

• RQ1 How can interactive time series visualization be made scalable to millions of data points without sacrificing responsiveness?
• RQ2 To what extent does dynamic data aggregation improve performance in real-time visual analytics of high-frequency time series?
• RQ3 Can a lightweight, extensible toolkit be built to integrate seamlessly into existing Python data science workflows like Jupyter and Dash?
• RQ4 How do different aggregation algorithms (e.g., EveryNth vs. LTTB) affect visual fidelity and aliasing artifacts in time series plots?
• RQ5 Can the toolkit serve as a platform for researching and comparing novel time series aggregation techniques?

Key findings

• Plotly-Resampler achieves sub-second response times during zoom and pan interactions even with datasets exceeding 1 million data points.
• Benchmarking shows Plotly-Resampler significantly outperforms alternative tools in terms of scalability across both number of samples and number of time series.
• The use of LTTB-based aggregation with visual indicators ([R] prefix and bin size estimate) effectively reduces aliasing artifacts compared to naive methods like EveryNth.
• The toolkit enables end-to-end visual analysis of complex, multivariate time series (e.g., polysomnographic EEG/EMG data) in a single interactive dashboard.
• Integration into existing workflows requires minimal code overhead—only wrapping a Plotly figure with FigureResampler and calling show_dash().
• The architecture supports future research into new aggregation techniques by exposing the underlying data processing pipeline in a modular way.