|
Activity Number:
|
443
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 2, 2016 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Risk Analysis
|
|
Abstract #319283
|
|
|
Title:
|
Application of Data Mining Techniques to Pesticide Risk Assessment
|
|
Author(s):
|
Ayona Chatterjee* and Arjun Panda and Jacob Holmab and Eric Suess
|
|
Companies:
|
California State University at East Bay and California State University at East Bay and California State University at East Bay and California State University at East Bay
|
|
Keywords:
|
Total Diet Study ;
Pesticide risk assessment ;
Naive Bayes ;
Data wrangling ;
Hierarchical Clustering
|
|
Abstract:
|
The Total Diet Study (TDS) provides data for Pesticide residue levels for numerous food that are commonly consumed by an average individual in the United States. Raw and summarized data are available from 1995 to 2014. The broad aim of this study is to assess the relationship between a food type and the amount of pesticide present. We perform data wrangling and exploratory data analysis for the analytical data from the TDS. Pesticide residue data are often left censored observations and positively skewed. A single food product has multiple pesticide residues though not all are present at a level higher than the limit of quantification (LQ). We use Naïve Bayes to classify different pesticides into significant contributor or not for each food type. To evaluate the classifier we split the data in to training and testing data sets and observe the error rates. Finally for the significant pesticides, we want to further identify the single largest contributor to residue levels for each food type. A Bayesian Hierarchical Clustering algorithm is applied to the data set to allow us to identify a single significant pesticide residue.
|
Authors who are presenting talks have a * after their name.
