Here is the Plan: So What is the Question?

In this action research study of my classroom of 5th grade mathematics students, I investigated their understanding of the mathematical operations by having them write problems to match given equations. I discovered that writing a story to match an equation does provide insight into a student’s understanding of mathematical concepts, however, reading comprehension is a factor in the understanding. Readers who struggle with comprehension do struggle with understanding and writing math story problems. The discussion that follows the writing of a math story problem and the solving of the written problems helps to strengthen the students’ mathematical abilities as well as their communication skills and confidence levels. Through my study, students learned that it was alright to make mistakes because the learning from those mistakes is what is important. As a result of this research, I plan to continue to have students write stories to match given equations as a source of information about student understanding. I will continue to give opportunities to revisit those written problems as a tool to increase students’ comprehension and communication skills, as well as their confidence.

Food Handling Practices, Knowledge and Beliefs of Families with Young Children Based on the Health Belief Model

Objective: To determine current food handling practices, knowledge and beliefs of primary food handlers with children 10 years old and the relationship between these components. Design: Surveys were developed based on FightBac!™ concepts and the Health Belief Model (HBM) construct. Participants: The majority of participants (n= 503) were females (67%), Caucasians (80%), aged between 30 to 49 years old (83%), had one or two children (83%), prepared meals all or most of the time (76%) and consumed meals away from home three times or less per week (66%). Analysis: Descriptive statistics and inferential statistics using Spearman’s rank correlation coefficient (rho) (p<0.05 and one-tail) and Chi-square were used to examine frequency and correlations. Results: Few participants reached the food safety objectives of Healthy People 2010 for safe food handling practices (79%). Mixed results were reported for perceived susceptibility. Only half of the participants (53-54%) reported high perceived severity for their children if they contracted food borne illness. Most participants were confident of their food handling practices for their children (91%) and would change their food handling practices if they or their family members previously experienced food poisoning (79%). Participants’ reasons for high self-efficacy were learning from their family and independently acquiring knowledge and skills from the media, internet or job. The three main barriers to safe food handling were insufficient time, lots of distractions and lack of control of the food handling practices of other people in the household. Participants preferred to use food safety information that is easy to understand, has scientific facts, causes feelings of health-threat and has lots of pictures or visuals. Participants demonstrate high levels of knowledge in certain areas of the FightBac!TM concepts but lacked knowledge in other areas. Knowledge and cues to action were most supportive of the HBM construct, while perceived susceptibility was least supportive of the HBM construct. Conclusion: Most participants demonstrate many areas to improve in their food handling practices, knowledge and beliefs. Adviser: Julie A. Albrecht

Active Learning to Maximize Area Under the ROC Curve

In active learning, a machine learning algorithmis given an unlabeled set of examples U, and is allowed to request labels for a relatively small subset of U to use for training. The goal is then to judiciously choose which examples in U to have labeled in order to optimize some performance criterion, e.g. classification accuracy. We study how active learning affects AUC. We examine two existing algorithms from the literature and present our own active learning algorithms designed to maximize the AUC of the hypothesis. One of our algorithms was consistently the top performer, and Closest Sampling from the literature often came in second behind it. When good posterior probability estimates were available, our heuristics were by far the best.