Affinity Propagation SMOTE approach for Imbalanced dataset used in Predicting Student at Risk of Low Performance

Author(s):  
B. Laureano Lanie
2019 ◽  
Vol 20 (3) ◽  
pp. 570-593 ◽  
Author(s):  
Vanessa R. Ralph ◽  
Scott E. Lewis

Students who score within the bottom quartile on cognitive measures of math aptitude have been identified as at-risk for low performance in chemistry courses, with less attention as to why such differential performance persists. At-risk students struggle most differentially on assessment items related to the mole concept and stoichiometry. An exploration as to the nature of the differential performance observed became of great interest as the assessment of these topics rarely progresses beyond multiplication or division, and at-risk students who achieved proficiency with the mole concept and stoichiometry had no noticeable gaps in academic chemistry performance when compared to students scoring in the top three quartiles of math aptitude. Thus, students in first-semester general chemistry were surveyed to describe their solution processes toward assessment items involving the mole concept and stoichiometry. Three hundred and forty-eight students responded to all survey prompts with 101 identified as at-risk. Findings suggest that while all students were observed to struggle in the conceptualization of the algorithms by which they execute solution processes, not-at-risk chemistry students were more likely to achieve correct answers via chemically implausible solution pathways. Rather than suggest the removal of assessment practices involving algorithmic, multiple-choice assessment on these topics, the implications include practical suggestions and opportunities for further research toward improving the equitability of measures used to assess proficiency with stoichiometry.


Author(s):  
Nicksson Ckayo Arrais de Freitas ◽  
Ticiana L. Coelho Da Silva ◽  
José Antônio Fernandes De Macêdo ◽  
Leopoldo Melo Júnioer

Deep learning has gained much popularity in the past years due to GPU advancements, cloud computing improvements, and its supremacy, considering the accuracy results when trained on massive datasets. As with machine learning, deep learning models may experience low performance when handled with imbalanced datasets. In this paper, we focus on the trajectory classification problem, and we examine deep learning techniques for coping with imbalanced class data. We extend a deep learning model, called DeepeST (Deep Learning for Sub-Trajectory classification), to predict the class or label for sub-trajectories from imbalanced datasets. DeepeST is the first deep learning model for trajectory classification that provides approaches for coping with imbalanced dataset problems from the authors' knowledge. In this paper, we perform the experiments with three real datasets from LBSN (Location-Based Social Network) trajectories to identify who is the user of a sub-trajectory (similar to the Trajectory-User Linking problem). We show that DeepeST outperforms other deep learning approaches from state-of-the-art concerning the accuracy, precision, recall, and F1-score.


2013 ◽  
Vol 5 (3) ◽  
pp. 464-467 ◽  
Author(s):  
Annette Visconti ◽  
Theodore Gaeta ◽  
Michael Cabezon ◽  
William Briggs ◽  
Matthew Pyle

Abstract Background Residents deemed at risk for low performance on standardized examinations require focused attention and remediation. Objective To determine whether a remediation program for residents identified as at risk for failure on the Emergency Medicine (EM) Written Board Examination is associated with improved outcomes. Intervention All residents in 8 classes of an EM 1–3 program were assessed using the In-Training Examination. Residents enrolled in the Focused Board Intervention (FBI) remediation program based on an absolute score on the EM 3 examination of <70% or a score more than 1 SD below the national mean on the EM 1 or 2 examination. Individualized education plans (IEPs) were created for residents in the FBI program, combining self-study audio review lectures with short-answer examinations. The association between first-time pass rate for the American Board of Emergency Medicine (ABEM) Written Qualifying Examination (WQE) and completion of all IEPs was examined using the χ2 test. Results Of the 64 residents graduating and sitting for the ABEM examination between 2000 and 2008, 26 (41%) were eligible for the program. Of these, 10 (38%) residents were compliant and had a first-time pass rate of 100%. The control group (12 residents who matched criteria but graduated before the FBI program was in place and 4 who were enrolled but failed to complete the program) had a 44% pass rate (7 of 16), which was significantly lower (χ2  =  8.6, P  =  .003). Conclusions The probability of passing the ABEM WQE on the first attempt was improved through the completion of a structured IEP.


1998 ◽  
Vol 29 (2) ◽  
pp. 109-116 ◽  
Author(s):  
Margie Gilbertson ◽  
Ronald K. Bramlett

The purpose of this study was to investigate informal phonological awareness measures as predictors of first-grade broad reading ability. Subjects were 91 former Head Start students who were administered standardized assessments of cognitive ability and receptive vocabulary, and informal phonological awareness measures during kindergarten and early first grade. Regression analyses indicated that three phonological awareness tasks, Invented Spelling, Categorization, and Blending, were the most predictive of standardized reading measures obtained at the end of first grade. Discriminant analyses indicated that these three phonological awareness tasks correctly identified at-risk students with 92% accuracy. Clinical use of a cutoff score for these measures is suggested, along with general intervention guidelines for practicing clinicians.


2020 ◽  
Vol 29 (3) ◽  
pp. 1283-1300
Author(s):  
Xigrid T. Soto ◽  
Andres Crucet-Choi ◽  
Howard Goldstein

Purpose Preschoolers' phonological awareness (PA) and alphabet knowledge (AK) skills are two of the strongest predictors of future reading. Despite evidence that providing at-risk preschoolers with timely emergent literacy interventions can prevent academic difficulties, there is a scarcity of research focusing on Latinx preschoolers who are dual language learners. Despite evidence of benefits of providing Latinxs with Spanish emergent literacy instruction, few studies include preschoolers. This study examined the effects of a supplemental Spanish PA and AK intervention on the dual emergent literacy skills of at-risk Latinx preschoolers. Method A multiple probe design across four units of instruction evaluated the effects of a Spanish supplemental emergent literacy intervention that explicitly facilitated generalizations to English. Four Latinx preschoolers with limited emergent literacy skills in Spanish and English participated in this study. Bilingual researchers delivered scripted lessons targeting PA and AK skills in individual or small groups for 12–17 weeks. Results Children made large gains as each PA skill was introduced into intervention and generalized the PA skills they learned from Spanish to English. They also improved their English initial sound identification skills, a phonemic awareness task, when instruction was delivered in Spanish but with English words. Children made small to moderate gains in their Spanish letter naming and letter–sound correspondence skills and in generalizing this knowledge to English. Conclusion These findings provide preliminary evidence Latinx preschoolers who are dual language learners benefit from emergent literacy instruction that promotes their bilingual and biliterate development.


2020 ◽  
Vol 63 (11) ◽  
pp. 3714-3726
Author(s):  
Sherine R. Tambyraja ◽  
Kelly Farquharson ◽  
Laura Justice

Purpose The purpose of this study was to determine the extent to which school-age children with speech sound disorder (SSD) exhibit concomitant reading difficulties and examine the extent to which phonological processing and speech production abilities are associated with increased likelihood of reading risks. Method Data were obtained from 120 kindergarten, first-grade, and second-grade children who were in receipt of school-based speech therapy services. Children were categorized as being “at risk” for reading difficulties if standardized scores on a word decoding measure were 1 SD or more from the mean. The selected predictors of reading risk included children's rapid automatized naming ability, phonological awareness (PA), and accuracy of speech sound production. Results Descriptive results indicated that just over 25% of children receiving school-based speech therapy for an SSD exhibited concomitant deficits in word decoding and that those exhibiting risk at the beginning of the school year were likely to continue to be at risk at the end of the school year. Results from a hierarchical logistic regression suggested that, after accounting for children's age, general language abilities, and socioeconomic status, both PA and speech sound production abilities were significantly associated with the likelihood of being classified as at risk. Conclusions School-age children with SSD are at increased risk for reading difficulties that are likely to persist throughout an academic year. The severity of phonological deficits, reflected by PA and speech output, may be important indicators of subsequent reading problems.


2020 ◽  
Vol 29 (4) ◽  
pp. 1944-1955 ◽  
Author(s):  
Maria Schwarz ◽  
Elizabeth C. Ward ◽  
Petrea Cornwell ◽  
Anne Coccetti ◽  
Pamela D'Netto ◽  
...  

Purpose The purpose of this study was to examine (a) the agreement between allied health assistants (AHAs) and speech-language pathologists (SLPs) when completing dysphagia screening for low-risk referrals and at-risk patients under a delegation model and (b) the operational impact of this delegation model. Method All AHAs worked in the adult acute inpatient settings across three hospitals and completed training and competency evaluation prior to conducting independent screening. Screening (pass/fail) was based on results from pre-screening exclusionary questions in combination with a water swallow test and the Eating Assessment Tool. To examine the agreement of AHAs' decision making with SLPs, AHAs ( n = 7) and SLPs ( n = 8) conducted an independent, simultaneous dysphagia screening on 51 adult inpatients classified as low-risk/at-risk referrals. To examine operational impact, AHAs independently completed screening on 48 low-risk/at-risk patients, with subsequent clinical swallow evaluation conducted by an SLP with patients who failed screening. Results Exact agreement between AHAs and SLPs on overall pass/fail screening criteria for the first 51 patients was 100%. Exact agreement for the two tools was 100% for the Eating Assessment Tool and 96% for the water swallow test. In the operational impact phase ( n = 48), 58% of patients failed AHA screening, with only 10% false positives on subjective SLP assessment and nil identified false negatives. Conclusion AHAs demonstrated the ability to reliably conduct dysphagia screening on a cohort of low-risk patients, with a low rate of false negatives. Data support high level of agreement and positive operational impact of using trained AHAs to perform dysphagia screening in low-risk patients.


2008 ◽  
Vol 18 (1) ◽  
pp. 24-31
Author(s):  
Martha Wilder Wilson ◽  
Elizabeth Zylla-Jones

Abstract The goal of university training programs is to educate speech-language pathology and audiology students to become competent and independent practitioners, with the ability to provide high quality and professional services to the public. This article describes the behaviors of “at-risk” student clinicians, so they may be identified early in their practica and remediation may be implemented. The importance of establishing a student at-risk protocol is discussed as well as a remediation plan for these students. This article summarized the Auburn University Speech and Hearing Clinic’s Student At-Risk Protocol, which may serve as a model for university training programs. The challenges of implementing such a protocol are also discussed.


ASHA Leader ◽  
2010 ◽  
Vol 15 (11) ◽  
pp. 1-38
Author(s):  
Marat Moore
Keyword(s):  

Sign in / Sign up

Export Citation Format

Share Document