User Tools

Site Tools


projects:year2:13.3

13.3 - Visual Analytic Approaches to Mining Large-Scale Time-Evolving Graphs

Project - Summary

Objectives:

The primary goal of this project is to develop a visual analytic framework for mining large-scale time evolving graphs. The development of the framework had three objectives: (1) we developed a methodology to construct a dependency graph using standard association analysis techniques to understand relationships between various entities (2) we developed a prediction model to predict event trends from evolutionary (or temporal) graphs, where individual nodes have non-stationary correlations, and (3) we investigated the application of emerging interaction and display devices for visual analytics interfaces.

Project - Team

Team Member Role Email Phone Number Academic Site/IAB
Raju Gottumukkala PI Not available Not available UL Lafayette
Christoph Borst Co-PI Not available Not available UL Lafayette
Siva Rama Krishna Venna Graduate Student Not available Not available UL Lafayette
Nicholas Lipari Graduate Student Not available Not available UL Lafayette

Project - Impact and Uses/Benefits

Using the symbolic time series to compute situational time lags: In state of the art or most existing techniques the time lag between the time series is computed using cross correlation between the numeric values. The main disadvantage of this process is that we are going to get a single time lag between these two time series. This may be suitable for some specific application like signal processing etc., where the time for impact or lag is constant. But in this case it is evident that flu spread occurs at different rates or with different time lags with varying climatic or weather conditions. Our proposed symbolic time series based time lag computation solved this problem by converting the normal numeric time series into symbolic or categorical time series where each category implies to a situation and then tries to find the appropriate time lags for each of the different categories. In this way we will be able to extract different time lags at different situations.

Two stage forecasting taking advantage of both Local and Global data: While some of the current flu forecasting or prediction systems use only the historical flu data to make forecasts or predictions and very few of these systems make use of the local climate or weather data for making the forecasts. In our proposed approach along with the local data for a city, we also use the global data as well. Here global data implies to the flu trends going on in this city’s surroundings or proximity. This is accomplished in this system by making a forecast first using the local data available and then in the next stage by observing and extracting the flu trends in the neighboring cities within the distance proximity, and making an adjustment to the forecast. The results prove the importance of this step.

Project - Deep Dive

Project - Documents

FilenameFilesizeLast modified
13.3_year_2_presentation.pptx45.3 MiB2019/08/22 11:13
13.3_year_2_ip_disclosure_letter.pdf1004.5 KiB2019/08/22 11:13
13.3_year2_final_project_report_combined.pdf5.1 MiB2019/08/22 10:12
projects/year2/13.3.txt · Last modified: 2019/08/22 10:14 by sally.johnson