This is a list of potential changes that can be made to the design based on results from the usablity test. Identify which categories below most closely matches with the users needs.
-
Data: show me a different metric
-
Comparison: show me a reference point
-
Trend: show me how things have changed over time
-
Outliers: show me data points that do not fit the norm
Outliers identify problem areas or data points that are doing unusually well. -
Distribution: show data values are spread across a scale.
Distribution is used to show the shape of data - composition of data values that makes that up the totals. Knowing the total numbers isn't enough in this case. -
Location: show me how things relate spatially
This shows more than just numbers. A visual representation of how things are mapped can provide insights. -
Part-to-whole relationship: show the set of values that contribute to a whole
-
Proportions: show relative contributions of several values to a whole
-
Patterns/Relationships: show correlations and associations among values
This reveals relationship between dimmensions/measures and how they affect each other.
-
breakout by right dimensions
-
Indexing
Index values to a common point.
Use this when the values have huge variations making it hard to view and compare points.
Use indexing to compare how values have changed. Indexing provides a common benchmark making it easier to track change. -
Scale
Start the scale from the minimum value in the dataset instead of zero.
Use this techinique when data points are hard to see because of a large variation in values. (Note that starting the scale from the minimum value will not work for all types of charts. A bar chart must always have a zero baseline. Dot plots and line charts are examples of charts that could start from a non zero baseline.) -
% vs. absolute values
Display the variance by converting absolute values to % change.
Use % if the user cares more about tracking change or growth rather than the actual values. -
Averages
Use mean as a measure of central tendency when the data set is uniform. A data set with extreme outliers will skew the mean. A median is a more accurate representation of the central tendency in this case. If there is a large difference between the median and mean then the distribution is probably skewed. Median should be used over mean in this case. Mean is best used for normal distributions. Mode can be used to measure the most common category. The problem with mode is that the most commonly occurring item might not be a good representation of the central tendency. -
Quartiles
Mark quartiles in your chart through bands of shaded color or lines. Use quartiles when the user wants additional context while viewing a disitribution. The lowest quartile will show where the lowest 25% of the values lie.
-
Filter: show me something conditionally
Add or remove filtering capability
Use fiters if the user is overwhelmed by too much information and wants to focus on one group at a time.
Filter can also be used for power users who want to be able to slice and dice the data is different ways. -
Brushing: show me connected items
Add or remove brushing capability
Use brushing when multiple visuizations on the screen have atleast one set of dimmensions that are the same. Brushing is useful when the user does not read charts in isolation and wants to view how a data point is doing across multiple visualizations. -
Reconfigure: show me a different arrangement
E.g. changing the sort criteria on a list
Reconfiguring can be used when the user does not have a fixed way of viewing information.
-
Encode: show me a different representation
Change the chart type altogether
Encoding is helpful when the user wants different insights from the same data points. -
Abstract/Elaborate: show me more or less detail
Add or remove tooltips on hover.
This can be used when the user wants additional context/details for only one set of information at a time. -
Select: mark something as interesting
Highlight points on the screen, add bookmarks Selections are used to highlight interesting elements on the screen and view them at a later timeor share them with someone else. -
Explore: show me something else
-
Alternative choices: show me alternative metrics/ dimensions
Add toggles for a visulization to switch out metrics Alternative choices are a great way to go back forth between two or more sets of metrics visualized the same way. -
Starring/tagging: show me things that I may have flagged
-
Annotate: show me my comments on a particular chart or page
-
Text-based summary: show me an automatically generated textual description of the key information in the dashboard
[Info Vis Wiki]: (http://www.infovis-wiki.net/index.php?title=Visualization_Design_Patterns)
Increase text size
Increase contrast between elements
Change colors to be more distinguishable
Meaningful Labels
Labels that describe what data is being displayed with enough business context
Is the chart's purpose clear from its title and subtitle?
Legends
Additional information: show me additional contextual information e.g. to learn more about how a metric is calculated/ how to read a chart
Visual cues
-
Placement: Where we locate things on a graph with attention to both position and proximity
-
Redundancy
Follow visual design principles to improve the overall aesthetics of your dashboard. https://www.oreilly.com/learning/visual-design-principles
Presentations on Visual Design Principles https://docs.google.com/presentation/d/182I3zWRQwAWXlLtB-DbfgeSrmC4XvVCpxz8sXzj7t-s/edit?usp=sharing https://docs.google.com/presentation/d/1slIHDPWSrEJEh-OikjZPDzuePkq8v7cd9Tfhq64wiqI/edit?usp=sharing
Flow Hierarchy Gradual Reveal
References: http://classes.engr.oregonstate.edu/eecs/spring2015/cs419-001/Slides/tufteDesign.pdf https://statistics.laerd.com/statistical-guides/measures-central-tendency-mean-mode-median.php https://www.perceptualedge.com/articles/visual_business_intelligence/introducing_bandlines.pdf