Project Team:

Lan Wu, wlmakeit@cs.ubc.ca

Project Consultant: Sherman Lai at UBC Fishery Department

Project Domain, Task, and Dataset:

Currently the prediction of impact on fishing is presented in form of a line chart with fish groups represented by different colors. However, due to the number of lines drawn on the screen, the chart suffers from an overload of visual information cluttering and overlapping. Thus, it is very difficult to retrieve any comprehensive information from the line chart system. See Figure 1. Groups of different fish species are interrelated though a diet matrix table that describes the amount of each group eats another group.

The project aims to optimize the line charts of fish population and visualize the information and trends in a more comprehensible way and give viewers a clear visual representation and interaction of the line chart.

The project is based on the data provided by the UBC Fishery Department.

Personal Expertise:

In terms of expertise, although Ying has taken an introductory biology course at UBC, and neither Ying nor Lan has any background in fishery, this seems like an interesting problem.

Proposed Infovis Solution:

• Line Layering with Brushing through th e Names of Fish Species. The idea of line layering is to differentiate fish species that are related via the diet matrix table from the ones that are not related. By moving the mouse onto the top of a line of a specific fish species, all the lines of its related fish species will be highlighted based on the diet matrix using a multiple color scheme. With respect to the highlighted lines, any other lines of unrelated fish species will be put in the background in gray color.
• Difference Graph. The idea of difference graph is to present the change of fish population before and after a fishing strategy is applied. The difference is computed between the set of data points of the current population of a fish species and that of the same fish population after a fishing strategy is applied. After the data set of difference is calculated, only the difference between the two populations is plotted on the graph.
• Line Clustering. The idea of line clustering is to group n similar lines of several fish species and present them using a new set of n number of lines to indicate a trend. There does not appear to be any existing line clustering library implemented; thus, an algorithm is proposed as the following:For each and every pair of lines, the average sum of squared difference between one line and another line will be calculated and store in an n x n matrix. Then given the number of clusters required, lines are grouped with respect to the smallest sum of squared difference to form a new line.In case the line clustering idea becomes unfeasible due to time constraint, the following optional solution will be proposed.


Optional Solutions:


• Line Chart Zooming. For the current line charts, all the lines drawn are on top of each other with very little deviations from one line to another. This causes extreme visual occlusion and overlapping. Zooming targets at this problem by enlarging a specific area of the line chart to a certain scale so that the difference among the lines can be easily viewed.
• Auto-Scrolling. The fish index list, Figure 1, contains the names of hundreds of marine species. Each time, only part of the list is visible in the window beside the line graph. Auto-scrolling targets at this problem by automatically scrolling down the list when the mouse reaches the end of the index window.

Usage Scenario:

• Fish Species Name Brushing with Layering Function. The user moves the mouse to the list of fish index and finds the fish's name 'Halibut'. As the fish's name is hovered over by the mouse, the lines of Halibut and its related fish species are highlighted with different color scheme, while all the lines of other unrelated ones are grayed out. As the user continuously moves the mouse to the name of another fish species, the same visualization keeps running until he withdraws the mouse from the fish index list
• Difference Graph Function. The user has already observed the highlighted lines and the grayed ones above. He also wants to see a difference graph showing what impact that a fishing strategy would have on the amount of a specific fish species. So he selects the line of that fish species on the line graph then moves the mouse underneath the line graph area and draws a fishing strategy curve. Afterwards he clicks the button named Difference Graph to have the difference line of the fish population change drawn in orange on the line graph with all other fish species lines grayed out.
• Line Clustering Function. Later in the analysis, the user wants to combine the population of several marine biomasses belonging to the same habitat location and see the trend. He multiply selects a few lines on the line graph and clicks the button named Clustering. All the lines of these selected fish species are clustered into one highlighted line and drawn in the line graph system.

Implementation Approach:

Project Milestones

• Nov 1, 2005 - User interview completed. The user interview with Sherman Lai at UBC Fishery Department helped us to get familiar with EcoPath and get to know what he wants to improve.
• Nov 4, 2005 - Proposal completed
• Nov 5-12, 2005 - Further test the feasibility of our proposal. Especially, we are not sure whether we can implement the line clustering function at this moment
• Nov 13, 2005 - Proposal update completed with updated milestones
• Nov 22, 2005 - Building of the software system EcoPath and design of classes.
• Nov 28, 2005 - Implimenting the line clustering function (or another one of optional components)
• Dec 4, 2005 - Implimenting the difference graph function
• Dec 10, 2005 - Implementating Fish Name Brushing with layering. Implementation phase completed
• Dec 14, 2005 - Testing.
• Dec 19, 2005 - Final presentation and final report completed