Statistics and Visualization of Point-Patterns

Stoev, Stilian; Hu, Weifeng

Work Description

Title: Statistics and Visualization of Point-Patterns Open Access Deposited

Attribute	Value
Methodology	The data was collected on the website Tornado History Project. The data is loaded into the program and visualized using Python heatmap.
Description	Many data sets come as point patterns of the form (longitude, latitude, time, magnitude). The examples of data sets in this format includes tornado events, origins/destination of internet flows, earthquakes, terrorist attacks and etc. It is difficult to visualize the data with simple plotting. This research project studies and implements non-parametric kernel smoothing in Python as a way of visualizing the intensity of point patterns in space and time. A two-dimensional grid M with size mx, my is used to store the calculation result for the kernel smoothing of each grid points. The heat-map in Python then uses the grid to plot the resulting images on a map where the resolution is determined by mx and my. The resulting images also depend on a spatial and a temporal smoothing parameters, which control the resolution (smoothness) of the figure. The Python code is applied to visualize over 56,000 tornado landings in the continental U.S. from the period 1950 - 2014. The magnitudes of the tornado are based on Fujita scale.
Creator	Stoev, Stilian Hu, Weifeng
Depositor	weifhu@umich.edu
Contact information	sstoev@umich.edu
Discipline	Science Engineering
Citations to related material	Hu, Weifeng. “Kernel-based Visualization of Point Patterns in Python with Application to Tornado Landing Data.” (2016). At https://www.semanticscholar.org/paper/Kernel-based-Visualization-of-Point-Patterns-in-to-Hu/0de06a6db39da54fe28f8d0cb47c0d3270f2f831
Resource type	Dataset
Last modified	02/24/2020
Published	04/24/2016
DOI	https://doi.org/10.7302/j00f-en55
License	http://creativecommons.org/licenses/by/3.0/us/

To Cite this Work:
Stoev, S., Hu, W. (2016). Statistics and Visualization of Point-Patterns [Data set]. University of Michigan - Deep Blue. https://doi.org/10.7302/j00f-en55

Files (Count: 4; Size: 2.75 MB)

Title	Original Upload	Last Modified	File Size	Access	Actions
READ_ME.txt	2017-04-05	2017-04-05	524 Bytes	Open Access	View Details Download
tornado_complete.txt	2017-04-05	2017-04-05	1.54 MB	Open Access	View Details Download
map_visualization_tor.py	2016-04-24	2016-04-24	8.06 KB	Open Access	View Details Download
Report.pdf	2017-04-05	2017-04-05	1.2 MB	Open Access	View Details Download

The dataset in "tornado_complete.txt" file is free for download on this website: http://www.tornadohistoryproject.com

The data are arranged in the following format with white spaces in between:

TouchdownLong(degree) TouchdownLat(degree) Fraction_of_year(unitless) Fujita_Scale(F)

To run the program, download both the source code as well as "tornado_complete.txt" file to the same directory. Run the program and enters number of pixels you want in both x and y directions for resolution, as well as bandwidth of time.

Download All Files (To download individual files, select them in the “Files” panel above)

Analytics

View Work Analytics