A Ranking in Teas


There is as much enjoyment in the process and eccentricities of brewing individual teas as there is in consuming the resulting cup. I wanted to be able to analyze and visualize what kinds of trends or overarching themes might exist in my personal tea consumption over a two month period. The goal of the visualization was to emphasize momentum, exploring how the brewing from one leaf might flow into another, and discover if there are any recurring patterns in the types of tea over time. Identifying the relationship between leafs can be a large factor in forming a more unified understanding of my tea perspective.

Interactive Visualization (View)

Graduate Coursework (SI649)

Fall 2012

Understanding the Data

A handwritten log was kept, tracking every record from September 10th to October 21st. The data collected contained information on the date and time of the infusion, the name of the leaf, the steep number, the pot used, and the temperature of the water.Together with information on the leaf origin and type, the data sets enabled a full inspection of tea habits through any pivot on the data that might prove useful in formalizing existent patterns across the leafs and time.

Approaching the data with the question intents in mind, the most important factor had to be maintaining a salience between the individual leafs. However, faced with the large number of otherwise categorical values, directly keeping the density from becoming overwhelming across time proved difficult. A certain level of abstraction was needed to clarify not precise activity of each leaf, but the relationship between different leafs concurrent activities.

Early mockup of the visualization, exploring how the different retinal variables would interact with the format.

The final interface, featuring the bumps chart, sorting controls, filtering by leaf, and an option to re-animate the chart. The visualization can be seen live here.

Visualization Techniques

The visualization technique employed borrows on what Edward Tufte refers to as a "bumps chart". Capable of showing place in time, the bumps chart depicts an ordered ranking between entities at a specific moment, and how the rankings change across a sequential set of discrete moments. For the current implementation the method is used reflect the graded popularity of each leaf and visualize the shifts as activity compounds across a data range.

The idea is to order the individual leaves, providing a ranking based on the total, cumulative infusions of each over the course of the time frame. The ranks will then represent the current state of which teas have been steeped the most, down to which have seen the least activity to date. Placing this as the primary mapping attribute allows an easy assessment of the complexity of changes that occur, and the relative reigns of each tea, without overwhelming the display. As we then add additional attributes, the value of comparisons between the leaf lines begins to form.

The highlight-by-type dropdown colorfully displays optinons for filtering the display down to a specific type of tea leaf.

Design Rationale

With few discrete properties, the presentation of the data had to deal with a very selective availability of valuable retinal variables. The decision made was to focus on the two most accessible variables to nominal and ordinal data: Position and Color. Using the horizontal axis as an ordinal mapping for the date, the most important characteristic, the leafs was then mapped to the remaining most most salient variable, the vertical axis. This allowed the rankings to alter vertically, allowing the orientational metaphor of "higher" on the list to correctly correlation to a greater degree of cumulative activity.

The choice of the chart model was made to best reflect the relationships of the leaves while not overwhelming the display. The "bumps chart" model proved very successful in achieving this while still carrying a substantial amount of data. In fact, the implementation of space and line results in a very high data-ink ratio, with almost all the visual details carrying some importance to the conveyance of the information.

The decision to include animation was based on concerns regarding the attentional capabilities of viewers. The ability to modify the date range was very much desirable, however the fast switching of the entire array of leaf lines could potentially cause the user to loose their orientation to the data display, and even result in instance of change blindness as attention moves between the date sliders and the dense curves. Slowly redrawing the progression of the curves alleviates some of this, both in leveraging the pre-attentive draw of movement, and forces an active attention to the changes in the lines as opposed to just the resultant endpoints.

Zooming in to a specific date range can allow a closer examination of when certian leafs stole the favorite position from others.