No abstract

A dropout early warning system enables schools to preemptively identify students who are at risk of dropping out of school, to promptly react to them, and eventually to help potential dropout students to continue their learning for a better future. However, the inherent class imbalance between dropout and non-dropout students could pose difficulty in building accurate predictive modeling for a dropout early warning system. The present study aimed to improve the performance of a dropout early warning system: (a) by addressing the class imbalance issue using the synthetic minority oversampling techniques (SMOTE) and the ensemble methods in machine learning; and (b) by evaluating the trained classifiers with both receiver operating characteristic (ROC) and precision–recall (PR) curves. To that end, we trained random forest, boosted decision tree, random forest with SMOTE, and boosted decision tree with SMOTE using the big data samples of the 165,715 high school students from the National Education Information System (NEIS) in South Korea. According to our ROC and PR curve analysis, boosted decision tree showed the optimal performance.

The rich data that Massive Open Online Courses (MOOCs) platforms collect on the behavior of millions of users provide a unique opportunity to study human learning and to develop data-driven methods that can address the needs of individual learners. This type of research falls into the emerging field of learning analytics. However, learning analytics research tends to ignore the issue of the reliability of results that are based on MOOCs data, which is typically noisy and generated by a largely anonymous crowd of learners. This paper provides evidence that learning analytics in MOOCs can be significantly biased by users who abuse the anonymity and open-nature of MOOCs, for example by setting up multiple accounts, due to their amount and aberrant behavior. We identify these users, denoted fake learners, using dedicated algorithms. The methodology for measuring the bias caused by fake learners' activity combines the ideas of Replication Research and Sensitivity Analysis. We replicate two highlycited learning analytics studies with and without fake learners data, and compare the results. While in one study, the results were relatively stable against fake learners, in the other, removing the fake learners' data significantly changed the results. These findings raise concerns regarding the reliability of learning analytics in MOOCs, and highlight the need to develop more robust, generalizable and verifiable research methods.

In vivo detection and quantification of cells labeled with superparamagnetic iron oxide (SPIO) nanoparticles has been attracting increasing attention. In particular, positive contrast methods, such as susceptibility gradient mapping (SGM) and phase gradient mapping (PGM), have been proposed for the improved detection of SPIO nanoparticles. In this study, a different implementation of the PGM method is introduced; it calculates the phase gradient in the image space using a fast Fourier transform without the need for phase unwrapping. We first compared positive contrast generation between the PGM and SGM methods, which estimates the susceptibility gradient in k space through echo shift measurements. Next, PGM was applied to quantify SPIO concentrations by fitting the resulting phase gradient maps to those of a theoretical model. MR experiments were conducted using a 3-T magnet scanner to acquire two datasets: the first was acquired from a gelatin phantom with three SPIO-doped vials of different concentrations, and the second was obtained in vivo from a nude rat with SPIO-labeled C6 glioma cells implanted in the flanks. The sensitivity of the PGM and SGM methods was compared using various factors, including different SPIO concentrations, TEs and signal-to-noise ratios. Based on the theoretical model of an infinite cylinder, the results demonstrated that, without loss of spatial resolution, the PGM method presents positive contrast maps with a higher sensitivity than SGM at medium and low SPIO concentrations, whereas SGM is more sensitive than PGM at longer TEs. The quantification of SPIO concentrations using the phantom dataset was also reported. On the basis of the same infinite cylinder model, it was shown that the PGM method provides an accurate estimation of SPIO concentration.

The present study focused on parents’ social cue use in relation to young children's attention. Participants were ten parent–child dyads; all children were 36 to 60 months old and were either typically developing (TD) or were diagnosed with autism spectrum disorder (ASD). Children wore a head-mounted camera that recorded the proximate child view while their parent played with them. The study compared the following between the TD and ASD groups: (a) frequency of parent's gesture use; (b) parents’ monitoring of their child's face; and (c) how children looked at parents’ gestures. Results from Bayesian estimation indicated that, compared to the TD group, parents of children with ASD produced more gestures, more closely monitored their children's faces, and provided more scaffolding for their children's visual experiences. Our findings suggest the importance of further investigating parents’ visual and gestural scaffolding as a potential developmental mechanism for children's early learning, including for children with ASD.