Home » News » Technology »

Student research could be used to search out, bust meth labs

 
Student research could be used to search out, bust meth labs
Graduate assistants Kenneth Robertson and Alexander Beregulko (not photographed) won first place in the national SAS M2010 Data Mining Conference for their project which uses data mining techniques to predict Methamphetamine Production In The Midwest. Robertson, Beregulko and supporting mathematics professor Carl Lee are currently working to publish the study in a leading peer review journal. “I was excited to extend my data mining possibilities,” Robertson said. (Paige Calamari/Staff Photographer)

Imagine creating a geo-spatial datamining map that can predict the occurrence of methamphetamine production sites based on the variables of human geography.

Kenneth Roberts did.

The Pinckney graduate student won the SAS Data Mining M 2010 student poster competition. He co-created a poster which visualized the most likely locations of meth labs based on statistics.

“The main reason I chose this project is because this is the only crime where there’s a longitude and latitude,” Roberts said. “Nationally any meth lab that’s found, the police make a record of it. I actually have an address for the methamphetamine labs, so from there I was able to get longitude and latitude location.”

Roberts processed the statistics through computer programs he created. He used an advanced data mining system called Geographic Information System (GIS) for hydraulic modeling and cartography.

“Just think Google Earth whenever you hear GIS,” he said.

Alexander Beregulko was the poster’s co-author. The Russia graduate student met Roberts in STA 591: Statistical Methods for Data Mining last spring. The two began working on a similar poster for a class project, which eventually led to the creation of the meth production poster.

The research concluded urban areas on the fringe of cities are more likely to have occurrences of meth labs.

“The results of my project were that we created a statistical model that predicts methamphetamine productions in the Midwest in accuracy of 87.5 percent,” Roberts said.

He said he hopes police will be able to utilize the material to better understand and stop crimes related to methamphetamine.

“We extended the frames of that project by adding data from other states and crime dates,” Beregulko said. “Ken mostly did GIS and I was working on the model, the mathematics of the map.”

They did their data research over the summer and created the poster when school began. Out of around 30 different schools, 5 winners were chosen. Roberts and Beregulko placed first.

As their prize, Roberts and Beregulko took an all expenses paid trip to Las Vegas. Their poster was on in Caesars Palace in Las Vegas at the 13th Annual Data Mining Conference in Caesar’s Palace.

“It felt great,” Beregulko said. “It was a relief that our efforts had been rewarded.”

Assistant Professor of Geography Brian Becker had Roberts in several classes. He said Roberts does his work quietly and is always one of the better students.

Becker credited Robert’s achievements to his ability to combine computer programming with geospatial research.

“The reason he’s so marketable is that he’s not just a GIS person, but he’s also a programmer,” Becker said. “That’s the niche he’s filled.”

Beregulko wants to find a job in financial analysis, he said it’s the first level you can work in the finance field

Roberts wants to work for either the National Geospatial-Intelligence Agency in Washington D.C. or in heath care predictive modeling.

“GIS and statistics were rated in the top 5 quality of life jobs by CNN Money last week,” he said. “It’s not mindless work. It’s a tool that can be used for sciences, so it furthers humanity.”

 
 
  • The Truth

    I wish there were more information here on how the model was made and what exactly it predicts. If it just extrapolates upon police reports to say that meth labs are usually on the fringes of cities, that's not very impressive. And if these predictions are taken seriously, they could contribute to even more police profiling of people in certain neighborhoods.

  • mae

    well they seem to find a lot of meth labs in rural settings, such as Harrison–breaking news this week another meth lab was busted.