We’re not the only cool graph in town

Every graph and plot has its place. Take a look at some fantastic alternatives to Jump Plot that may fit your needs when analyzing distributions.

Box Plot

  • Highlights distribution quartiles more clearly.
  • Can highlight statistical outliers by design.
  • Variations allow for specific step information to the level desired.
  • If used for event sequence, there is a limitation where all hops must be met for the distribution values to be meaningful.
  • Otherwise you may have steps skipped and cannot discern from visual alone (with addition of z axis).
  • Box plot uses dots and boxes which breaks up the visual in terms of data over time. Think of bar chart used for time analysis instead of line chart.
  • Ability for hops to skip steps in sequence without affecting the distribution visualization.
  • Jump plot visualization is more easily interpreted as time based due to the left-to-right nature of the graph design.
  • Distribution of measures (although Jump Plot focuses on time based measures more than box plot).
  • Threshold based review (in both cases threshold would placed as the horizontal center of distribution).
  • Percentage transformation can be applied to both.
  • Reference lines can be applied to both

Directed Graph

  • Its dynamic nature of node placement may allow for easier review of sequences with high checkpoint counts (i.e., checkpoint scaling).
  • The “hairball” style of the Directed Graph does not lend itself to distribution analysis of “hop” distributions between checkpoints.
  • For time analysis, the preferred visualization is from left to right (e.g., line graph). With the Directed Graph nodes may not be placed in this linear left to right view. Readers will have to spend more time review to quantify time with the directed graph.
  • Overplotting will occur more frequently due to “hairball” style of directed graph.
  • Linear, left-to-right view, synonymous to readers as duration over time.
  • Better ability to evaluate distributions between checkpoints due to linear nature.
  • Less overplotting
  • Easier interpretation/analysis of:
    • Percentage Transformation
    • Threshold Basis
    • Panel/Trellis VIew
    • Reference Lines
  • Network of checkpoints and associated hops (with/without hop skips) in between.

Line Chart

  • Visualization purist view of events over time (Jump Plot may use both axes to view time).
  • Without overplotting, it can only show a single measure per checkpoint (e.g., avg checkpoint duration).
  • Ability to review distributions across multiple checkpoint
  • Less overplotting
  • Easier to review various series distributions through checkpoints
  • Easier to identify checkpoints that have been skipped
  • Easier interpretation/analysis of:
    • Percentage Transformation
    • Distribution of multiple series (when series have multiple pass throughs).
  • Time based view (e.g., linear, left-to-right).
  • Reference Lines
  • Threshold Based
  • Panel/Trellis View
  • Series (if series only contain a single pass through of checkpoints or is aggregated).

Time Geography

  • Thoughts on “Time Geography” compared to Jump Plot
    • Not linear (left-to-right) thus does not lend itself to time from reader’s perspective.
    • Have to interpret the day up and down in terms of a gantt chart.
    • Distribution of times over high volume series events would lead to a great deal of over-plotting or scrolling required.
    • Time Geography visualization allows for a checkpoint/event to be re-visited within a sequence events, this is not a strength of the Jump Plot, but can be addressed within the Jump Plot.
    • Reference lines could potentially be applied.
    • Percentage transformation would be hard to implement.
    • Threshold based analysis would be tough.


  • Thoughts on “Lifeflow” compared to Jump Plot
    • Does show activity from left to right.
    • Does not lend itself to distribution between checkpoints.
    • More like a decision tree of what volume flows from checkpoint 1 to 2, etc.
    • Series are graphed as separate rows from top to bottom, makes it harder to interpret distributions between checkpoints (have to scan for section of horizontal axis from top to bottom).
    • Has similar issue to area chart, you can only view one step in the sequence at a time (by resetting the primary axis for each checkpoint).
    • Requires z-axis for initial review of event sequence, thus limits the ability for additional series analysis.