Evaluation of sampling methods for learning from imbalanced data

The problem of learning from imbalanced data is of critical importance in a large number of application domains and can be a bottleneck in the performance of various conventional learning methods that assume the data distribution to be balanced. The class imbalance problem corresponds to dealing with the situation where one class massively outnumbers the other. The imbalance between majority and minority would lead machine learning to be biased and produce unreliable outcomes if the imbalanced data is used directly. There has been increasing interest in this research area and a number of algorithms have been developed. However, independent evaluation of the algorithms is limited. This paper aims at evaluating the performance of five representative data sampling methods namely SMOTE, ADASYN, BorderlineSMOTE, SMOTETomek and RUSBoost that deal with class imbalance problems. A comparative study is conducted and the performance of each method is critically analysed in terms of assessment metrics. © 2013 Springer-Verlag.

The Complexity of Approximately Solving Influence Diagrams

Influence diagrams allow for intuitive and yet precise description of complex situations involving decision making under uncertainty. Unfortunately, most of the problems described by influence diagrams are hard to solve. In this paper we discuss the complexity of approximately solving influence diagrams. We do not assume no-forgetting or regularity, which makes the class of problems we address very broad. Remarkably, we show that when both the treewidth and the cardinality of the variables are bounded the problem admits a fully polynomial-time approximation scheme.

Discrete Conditional Phase-type Model Utilising a Multiclass Support Vector Machine for the Prediction of Retinopathy of Prematurity

Retinopathy of prematurity (ROP) is a rare disease in which retinal blood vessels of premature infants fail to develop normally, and is one of the major causes of childhood blindness throughout the world. The Discrete Conditional Phase-type (DC-Ph) model consists of two components, the conditional component measuring the inter-relationships between covariates and the survival component which models the survival distribution using a Coxian phase-type distribution. This paper expands the DC-Ph models by introducing a support vector machine (SVM), in the role of the conditional component. The SVM is capable of classifying multiple outcomes and is used to identify the infant's risk of developing ROP. Class imbalance makes predicting rare events difficult. A new class decomposition technique, which deals with the problem of multiclass imbalance, is introduced. Based on the SVM classification, the length of stay in the neonatal ward is modelled using a 5, 8 or 9 phase Coxian distribution.

A cephalometric study of class II malocclusions treated with mandibular surgery

INTRODUCTION:

Class II malocclusion is often associated with retrognathic mandible. Some of these problems require surgical correction. The purposes of this study were to investigate treatment outcomes in patients with Class II malocclusions whose treatment included mandibular advancement surgery and to identify predictors of good outcomes.
METHODS:

Pretreatment and posttreatment cephalometric radiographs of 90 patients treated with mandibular advancement surgery by 57 consultant orthodontists in the United Kingdom before September 1998 were digitized, and cephalometric landmarks were identified. Paired samples t tests were used to compare the pretreatment and posttreatment cephalometric values for each patient. For each cephalometric variable, the proportion of patients falling within the ideal range was identified. Multiple logistic regression analysis was performed to identify predictors of achieving ideal range outcomes for the key skeletal (ANB and SNB angles), dental (overjet and overbite), and soft-tissue (Holdaway angle) measurements.
RESULTS:

An overjet within the ideal range of 1 to 4 mm was achieved in 72% of patients and was more likely with larger initial ANB angles. Horizontal correction of the incisor relationship was achieved by a combination of 75% skeletal movement and 25% dentoalveolar change. An ideal posttreatment ANB angle was achieved in 42% of patients and was more likely in females and those with larger pretreatment ANB angles. Ideal soft-tissue Holdaway angles (7 degrees to 14 degrees ) were achieved in 49% of patients and were more likely in females and those with smaller initial SNA angles. Mandibular incisor decompensation was incomplete in 28% of patients and was more likely in females and patients with greater pretreatment mandibular incisor proclination. Correction of increased overbite was generally successful, although anterior open bites were found in 16% of patients at the end of treatment. These patients were more likely to have had initial open bites.
CONCLUSIONS:

Mandibular surgery had a good success rate in normalizing the main dental and skeletal relationships. Less ideal soft-tissue profile outcomes were associated with larger pretreatment SNA-angle values, larger final mandibular incisor inclinations, and smaller final maxillary incisor inclinations. The use of mandibular surgery to correct anterior open bite was associated with poor outcomes.

The validity of a latent class typology of adolescent drinking patterns

Objectives: This study examined the validity of a latent class typology of adolescent drinking based on four alcohol dimensions; frequency of drinking, quantity consumed, frequency of binge drinking and the number of alcohol related problems encountered. Method: Data used were from the 1970 British Cohort Study sixteen-year-old follow-up. Partial or complete responses to the selected alcohol measures were provided by 6,516 cohort members. The data were collected via a series of postal questionnaires. Results: A five class LCA typology was constructed. Around 12% of the sample were classified as �hazardous drinkers� reporting frequent drinking, high levels of alcohol consumed, frequent binge drinking and multiple alcohol related problems. Multinomial logistic regression, with multiple imputation for missing data, was used to assess the covariates of adolescent drinking patterns. Hazardous drinking was associated with being white, being male, having heavy drinking parents (in particular fathers), smoking, illicit drug use, and minor and violent offending behaviour. Non-significant associations were found between drinking patterns and general mental health and attention deficient disorder. Conclusion: The latent class typology exhibited concurrent validity in terms of its ability to distinguish respondents across a number of alcohol and non-alcohol indicators. Notwithstanding a number of limitations, latent class analysis offers an alternative data reduction method for the construction of drinking typologies that addresses known weaknesses inherent in more tradition classification methods.

Understanding the 'production' of the schooling identities of young working class boys: time to move beyond the sociological/psychological divide?

In recent times the sociology of childhood has played an important role in challenging the dominance of Piagetian models of child development in shaping the way we think about children and childhood. What such work has successfully achieved is to increase our understanding of the socially constructed nature of childhood; the social competence and agency of children; and the diverse nature of children’s lives, reflecting the very different social contexts within which they are located. One of the problems that has tended to be associated with this work, however, is that in its critique of developmentalism it has tended simply to replace one orthodoxy (psychology) with another (sociology) rather than providing the opportunity to transcend this divide. The purpose of this paper is to demonstrate some of the potential ways in which the sociological/psychological divide might be transcended and the benefits of this for understanding, more fully, the ‘production’ of children’s schooling identities. In particular it shows how some of the key sociological insights to be found in the work of Bourdieu may be usefully extended by the work inspired by the developmental psychologist, Vygotsky. The key arguments are illustrated by reference to ethnographic data relating to the schooling experiences and identities of a group of 5-6 year old working class boys.

Higher effort-reward imbalance and lower job control predict exit from the labour market at the age of 61 years or younger: Evidence from the English Longitudinal Study of Ageing

Background: We examined whether higher effort-reward imbalance (ERI) and lower job control are associated with exit from the labour market. 

Methods: There were 1263 participants aged 50-74 years from the English Longitudinal Study on Ageing with data on working status and work-related psychosocial factors at baseline (wave 2; 2004-2005), and working status at follow-up (wave 5; 2010-2011). Psychosocial factors at work were assessed using a short validated version of ERI and job control. An allostatic load index was formed using 13 biological parameters. Depressive symptoms were measured using the Center for Epidemiologic Studies Depression Scale. Exit from the labour market was defined as not working in the labour market when 61 years old or younger in 2010-2011. 

Results: Higher ERI OR=1.62 (95% CI 1.01 to 2.61, p=0.048) predicted exit from the labour market independent of age, sex, education, occupational class, allostatic load and depression. Job control OR=0.60 (95% CI 0.42 to 0.85, p=0.004) was associated with exit from the labour market independent of age, sex, education, occupation and depression. The association of higher effort OR=1.32 (95% CI 1.01 to 1.73, p=0.045) with exit from the labour market was independent of age, sex and depression but attenuated to non-significance when additionally controlling for socioeconomic measures. Reward was not related to exit from the labour market. 

Conclusions: Stressful work conditions can be a risk for exiting the labour market before the age of 61 years. Neither socioeconomic position nor allostatic load and depressive symptoms seem to explain this association.

A class of efficient short-chain fluorinated catanionic surfactants

There have been many attempts to find suitable replacements for fluorinated surfactants due to problems related to their bioaccumulation and resistance to biodegradation. Meeting the exceptional performances of these compounds, however, remains largely an unaccomplished challenge. Currently, a solution might be found through the synthesis of new classes of fluorinated compounds that possess reduced environmental impact by following the recommended strategies for their greener and safer design. In this article we report a novel approach by designing a family of catanionic fluorinated surfactants that shows potential for improved degradation, good water solubility, and low Krafft points. Furthermore, these surfactants exhibit excellent fluorine efficiency and effectiveness of surface tension reduction.

Transitions in Contentious Behaviour in Infancy and Later Aggressive Conduct Problems

Background. Individual trajectories toward aggression originate in early infancy, before there is intent to harm. We focused on infants who were contentious, i.e., prone to engage in anger and use of physical force with other people, and examined change in levels of contentiousness between 6 and 12 months of age with reference to later aggressive conduct problems.
Sample. The CCDS is a nationally representative sample of 321 firstborn children whose families were recruited from antenatal clinics in two National Health Service Trusts.
Method. Mothers, fathers, and a third family member or friend who knew infants well completed the Cardiff Infant Contentiousness Scale (CICS) at 6 months, which was stable form 6 to 12 months, and validated by direct observation of infants’ use of force against peers. Primary caregivers again completed the CICS at 12 months, and up to three informants completed the Child Behaviour Check List at mean ages of 36 and 84 months. We used Latent Transition Analysis to identify different groups of infants in respect to their patterns of contentiousness from 6 to 12 months.
Results
Three ordered classes of contentiousness from low to high were found at 6 and 12 months. Infants exposed to greater family adversity were more likely to move into the high-contentious class from 6 to 12 months. Higher contentiousness in infancy predicted more aggressive conduct problems at 33 months and thereafter.
Conclusions
Infants exposed to family adversity are already at disadvantage by 6 months and likely to escalate in their anger and aggressiveness over time.