GR1 - User and Task Analysis

User Analysis

Our representative users are taken from industry (Boingo, IBM), government (DOD), and academia (MIT). In general, users want an interface that is intuitive, interactive, and efficient. Users depend on data visualization to draw conclusions from complex data and make informed, time-sensitive decisions, so safety is an important usability aspect. Because users will generally become very familiar with the software in their day-to-day work, we will choose efficiency over learnability whenever necessary.

Industry: IBM is seeking an improved user interface for Analyst's Notebook, one that can compete with the more user-friendly Palantir software. Boingo needs to analyze data to build its new WiFi network in Africa, and needs a good way to visualize a combination of data sets across the continent.

Government: We will be collaborating with U.S. Army officers who use both Palantir and Analyst's Notebook for mission-critical decision making. One of the problems they face is the inefficiency of using these two different tools, as their supporting databases are incompatible. They wish to use the superior computational tools of Analyst's Notebook, but have an interface that can more clearly and efficiently visualize data trends.

Academia: Data analysis is an important part of research in many fields. In particular, research about international development in Africa has clear ties to the above mentioned users in industry and government.

One important design consideration is that different users want to see different levels of granularity in the data. For example, an Army major would want to see more technical details while a colonel may be more concerned with qualitative trends; similarly with an engineer versus a top executive in industry. In academia, researchers may want to look at data on many levels to draw connections between data trends.

Task Analysis

With the back-end and database support provided by IBM, we will focus on designing effective ways to visualize data in a way that is understandable to users. Standard data visualization tools are available from open-source projects such as processing.js. We will select a few visualization methods (e.g. web, map, graph) to focus on, geared toward the data sets for WiFi in Africa. Some high-level tasks include:

1. Creating visualizations: It should be easy for users to generate new visualizations, e.g. selecting best visualization for data set, selecting data sets for different axes, etc.
2. Interacting with visualizations: Analysts may want to interact with the visualization, e.g. sliding timescale, zoom in/out, etc.
   1. Commenting: We will look for ways to allow them to insert text comments, highlight parts of the visualization, etc. We will be in communication with the representative users to pinpoint the features that will be most useful to them.
3. Combining visualizations: Whether overlaying data sets or seeing different data sets next to each other, users may want to see multiple visualizations at once. This is one advantage that Palantir currently has over Analyst's Notebook.