skip to main content


Title: The NAEP EDM Competition: On the Value of Theory-Driven Psychometrics and Machine Learning for Predictions Based on Log Data
The 2nd Annual WPI-UMASS-UPENN EDM Data Min- ing Challenge required contestants to predict efficient test taking based on log data. In this paper, we describe our theory-driven and psychometric modeling approach. For feature engineering, we employed the Log-Normal Response Time Model for estimating latent person speed, and the Generalized Partial Credit Model for estimating latent person ability. Additionally, we adopted an n-gram feature approach for event sequences. For training a multi-label classifier, we distinguished inefficent test takers who were going too fast and those who were going too slow, instead of using the provided binary target label. Our best-performing ensemble classify er comprised three sets of low-dimensional classi ers, dominated by test-taker speed. While our classi- er reached moderate performance, relative to competition leaderboard, our approach makes two important contributions. First, we show how explainable classi ers could provide meaningful predictions if results can be contextualized to test administrators who wish to intervene or take action. Second, our re-engineering of test scores enabled us to incorporate person ability into the estimation. However, ability was hardly predictive of efficient behavior, leading to the conclusion that the target label's validity needs to be questioned. The paper concludes with tools that are helpful for substantively meaningful log data mining.  more » « less
Award ID(s):
1636847
NSF-PAR ID:
10190399
Author(s) / Creator(s):
; ; ; ; ; ;
Date Published:
Journal Name:
Educational Data Mining
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. null (Ed.)
    COVID-19 has altered the landscape of teaching and learning. For those in in-service teacher education, workshops have been suspended causing programs to adapt their professional development to a virtual space to avoid indefinite postponement or cancellation. This paradigm shift in the way we conduct learning experiences creates several logistical and pedagogical challenges but also presents an important opportunity to conduct research about how learning happens in these new environments. This paper describes the approach we took to conduct research in a series of virtual workshops aimed at teaching rural elementary teachers about engineering practices and how to teach a unit from an engineering curriculum. Our work explores how engineering concepts and practices are socially constructed through interactions with teachers, students, and artifacts. This approach, called interactional ethnography has been used by the authors and others to learn about engineering teaching and learning in precollege classrooms. The approach relies on collecting data during instruction, such as video and audio recordings, interviews, and artifacts such as journal entries and photos of physical designs. Findings are triangulated by analyzing these data sources. This methodology was going to be applied in an in-person engineering education workshop for rural elementary teachers, however the pandemic forced us to conduct the workshops remotely. Teachers, working in pairs, were sent workshop supplies, and worked together during the training series that took place over Zoom over four days for four hours each session. The paper describes how we collected video and audio of teachers and the facilitators both in whole group and in breakout rooms. Class materials and submissions of photos and evaluations were managed using Google Classroom. Teachers took photos of their work and scanned written materials and submitted them all by email. Slide decks were shared by the users and their group responses were collected in real time. Workshop evaluations were collected after each meeting using Google Forms. Evaluation data suggest that the teachers were engaged by the experience, learned significantly about engineering concepts and the knowledge-producing practices of engineers, and feel confident about applying engineering activities in their classrooms. This methodology should be of interest to the membership for three distinct reasons. First, remote instruction is a reality in the near-term but will likely persist in some form. Although many of us prefer to teach in person, remote learning allows us to reach many more participants, including those living in remote and rural areas who cannot easily attend in-person sessions with engineering educators, so it benefits the field to learn how to teach effectively in this way. Second, it describes an emerging approach to engineering education research. Interactional ethnography has been applied in precollege classrooms, but this paper demonstrates how it can also be used in teacher professional development contexts. Third, based on our application of interactional ethnography to an education setting, readers will learn specifically about how to use online collaborative software and how to collect and organize data sources for research purposes. 
    more » « less
  2. We consider the problem of estimating how well a model class is capable of fitting a distribution of labeled data. We show that it is possible to accurately estimate this “learnability” even when given an amount of data that is too small to reliably learn any accurate model. Our first result applies to the setting where the data is drawn from a d-dimensional distribution with isotropic covariance, and the label of each datapoint is an arbitrary noisy function of the datapoint. In this setting, we show that with O(sqrt(d)) samples, one can accurately estimate the fraction of the variance of the label that can be explained via the best linear function of the data. For comparison, even if the labels are noiseless linear functions of the data, a sample size linear in the dimension, d, is required to learn any function correlated with the underlying model. Our estimation approach also applies to the setting where the data distribution has an (unknown) arbitrary covariance matrix, allowing these techniques to be applied to settings where the model class consists of a linear function applied to a nonlinear embedding of the data. In this setting we give a consistent estimator of the fraction of explainable variance that uses o(d) samples. Finally, our techniques also extend to the setting of binary classification, where we obtain analogous results under the logistic model, for estimating the classification accuracy of the best linear classifier. We demonstrate the practical viability of our approaches on synthetic and real data. This ability to estimate the explanatory value of a set of features (or dataset), even in the regime in which there is too little data to realize that explanatory value, may be relevant to the scientific and industrial settings for which data collection is expensive and there are many potentially relevant feature sets that could be collected. 
    more » « less
  3. Abstract Purpose The ability to identify the scholarship of individual authors is essential for performance evaluation. A number of factors hinder this endeavor. Common and similarly spelled surnames make it difficult to isolate the scholarship of individual authors indexed on large databases. Variations in name spelling of individual scholars further complicates matters. Common family names in scientific powerhouses like China make it problematic to distinguish between authors possessing ubiquitous and/or anglicized surnames (as well as the same or similar first names). The assignment of unique author identifiers provides a major step toward resolving these difficulties. We maintain, however, that in and of themselves, author identifiers are not sufficient to fully address the author uncertainty problem. In this study we build on the author identifier approach by considering commonalities in fielded data between authors containing the same surname and first initial of their first name. We illustrate our approach using three case studies. Design/methodology/approach The approach we advance in this study is based on commonalities among fielded data in search results. We cast a broad initial net—i.e., a Web of Science (WOS) search for a given author’s last name, followed by a comma, followed by the first initial of his or her first name (e.g., a search for ‘John Doe’ would assume the form: ‘Doe, J’). Results for this search typically contain all of the scholarship legitimately belonging to this author in the given database (i.e., all of his or her true positives), along with a large amount of noise, or scholarship not belonging to this author (i.e., a large number of false positives). From this corpus we proceed to iteratively weed out false positives and retain true positives. Author identifiers provide a good starting point—e.g., if ‘Doe, J’ and ‘Doe, John’ share the same author identifier, this would be sufficient for us to conclude these are one and the same individual. We find email addresses similarly adequate—e.g., if two author names which share the same surname and same first initial have an email address in common, we conclude these authors are the same person. Author identifier and email address data is not always available, however. When this occurs, other fields are used to address the author uncertainty problem. Commonalities among author data other than unique identifiers and email addresses is less conclusive for name consolidation purposes. For example, if ‘Doe, John’ and ‘Doe, J’ have an affiliation in common, do we conclude that these names belong the same person? They may or may not; affiliations have employed two or more faculty members sharing the same last and first initial. Similarly, it’s conceivable that two individuals with the same last name and first initial publish in the same journal, publish with the same co-authors, and/or cite the same references. Should we then ignore commonalities among these fields and conclude they’re too imprecise for name consolidation purposes? It is our position that such commonalities are indeed valuable for addressing the author uncertainty problem, but more so when used in combination. Our approach makes use of automation as well as manual inspection, relying initially on author identifiers, then commonalities among fielded data other than author identifiers, and finally manual verification. To achieve name consolidation independent of author identifier matches, we have developed a procedure that is used with bibliometric software called VantagePoint (see www.thevantagepoint.com) While the application of our technique does not exclusively depend on VantagePoint, it is the software we find most efficient in this study. The script we developed to implement this procedure is designed to implement our name disambiguation procedure in a way that significantly reduces manual effort on the user’s part. Those who seek to replicate our procedure independent of VantagePoint can do so by manually following the method we outline, but we note that the manual application of our procedure takes a significant amount of time and effort, especially when working with larger datasets. Our script begins by prompting the user for a surname and a first initial (for any author of interest). It then prompts the user to select a WOS field on which to consolidate author names. After this the user is prompted to point to the name of the authors field, and finally asked to identify a specific author name (referred to by the script as the primary author) within this field whom the user knows to be a true positive (a suggested approach is to point to an author name associated with one of the records that has the author’s ORCID iD or email address attached to it). The script proceeds to identify and combine all author names sharing the primary author’s surname and first initial of his or her first name who share commonalities in the WOS field on which the user was prompted to consolidate author names. This typically results in significant reduction in the initial dataset size. After the procedure completes the user is usually left with a much smaller (and more manageable) dataset to manually inspect (and/or apply additional name disambiguation techniques to). Research limitations Match field coverage can be an issue. When field coverage is paltry dataset reduction is not as significant, which results in more manual inspection on the user’s part. Our procedure doesn’t lend itself to scholars who have had a legal family name change (after marriage, for example). Moreover, the technique we advance is (sometimes, but not always) likely to have a difficult time dealing with scholars who have changed careers or fields dramatically, as well as scholars whose work is highly interdisciplinary. Practical implications The procedure we advance has the ability to save a significant amount of time and effort for individuals engaged in name disambiguation research, especially when the name under consideration is a more common family name. It is more effective when match field coverage is high and a number of match fields exist. Originality/value Once again, the procedure we advance has the ability to save a significant amount of time and effort for individuals engaged in name disambiguation research. It combines preexisting with more recent approaches, harnessing the benefits of both. Findings Our study applies the name disambiguation procedure we advance to three case studies. Ideal match fields are not the same for each of our case studies. We find that match field effectiveness is in large part a function of field coverage. Comparing original dataset size, the timeframe analyzed for each case study is not the same, nor are the subject areas in which they publish. Our procedure is more effective when applied to our third case study, both in terms of list reduction and 100% retention of true positives. We attribute this to excellent match field coverage, and especially in more specific match fields, as well as having a more modest/manageable number of publications. While machine learning is considered authoritative by many, we do not see it as practical or replicable. The procedure advanced herein is both practical, replicable and relatively user friendly. It might be categorized into a space between ORCID and machine learning. Machine learning approaches typically look for commonalities among citation data, which is not always available, structured or easy to work with. The procedure we advance is intended to be applied across numerous fields in a dataset of interest (e.g. emails, coauthors, affiliations, etc.), resulting in multiple rounds of reduction. Results indicate that effective match fields include author identifiers, emails, source titles, co-authors and ISSNs. While the script we present is not likely to result in a dataset consisting solely of true positives (at least for more common surnames), it does significantly reduce manual effort on the user’s part. Dataset reduction (after our procedure is applied) is in large part a function of (a) field availability and (b) field coverage. 
    more » « less
  4. Abstract This project is funded by the US National Science Foundation (NSF) through their NSF RAPID program under the title “Modeling Corona Spread Using Big Data Analytics.” The project is a joint effort between the Department of Computer & Electrical Engineering and Computer Science at FAU and a research group from LexisNexis Risk Solutions. The novel coronavirus Covid-19 originated in China in early December 2019 and has rapidly spread to many countries around the globe, with the number of confirmed cases increasing every day. Covid-19 is officially a pandemic. It is a novel infection with serious clinical manifestations, including death, and it has reached at least 124 countries and territories. Although the ultimate course and impact of Covid-19 are uncertain, it is not merely possible but likely that the disease will produce enough severe illness to overwhelm the worldwide health care infrastructure. Emerging viral pandemics can place extraordinary and sustained demands on public health and health systems and on providers of essential community services. Modeling the Covid-19 pandemic spread is challenging. But there are data that can be used to project resource demands. Estimates of the reproductive number (R) of SARS-CoV-2 show that at the beginning of the epidemic, each infected person spreads the virus to at least two others, on average (Emanuel et al. in N Engl J Med. 2020, Livingston and Bucher in JAMA 323(14):1335, 2020). A conservatively low estimate is that 5 % of the population could become infected within 3 months. Preliminary data from China and Italy regarding the distribution of case severity and fatality vary widely (Wu and McGoogan in JAMA 323(13):1239–42, 2020). A recent large-scale analysis from China suggests that 80 % of those infected either are asymptomatic or have mild symptoms; a finding that implies that demand for advanced medical services might apply to only 20 % of the total infected. Of patients infected with Covid-19, about 15 % have severe illness and 5 % have critical illness (Emanuel et al. in N Engl J Med. 2020). Overall, mortality ranges from 0.25 % to as high as 3.0 % (Emanuel et al. in N Engl J Med. 2020, Wilson et al. in Emerg Infect Dis 26(6):1339, 2020). Case fatality rates are much higher for vulnerable populations, such as persons over the age of 80 years (> 14 %) and those with coexisting conditions (10 % for those with cardiovascular disease and 7 % for those with diabetes) (Emanuel et al. in N Engl J Med. 2020). Overall, Covid-19 is substantially deadlier than seasonal influenza, which has a mortality of roughly 0.1 %. Public health efforts depend heavily on predicting how diseases such as those caused by Covid-19 spread across the globe. During the early days of a new outbreak, when reliable data are still scarce, researchers turn to mathematical models that can predict where people who could be infected are going and how likely they are to bring the disease with them. These computational methods use known statistical equations that calculate the probability of individuals transmitting the illness. Modern computational power allows these models to quickly incorporate multiple inputs, such as a given disease’s ability to pass from person to person and the movement patterns of potentially infected people traveling by air and land. This process sometimes involves making assumptions about unknown factors, such as an individual’s exact travel pattern. By plugging in different possible versions of each input, however, researchers can update the models as new information becomes available and compare their results to observed patterns for the illness. In this paper we describe the development a model of Corona spread by using innovative big data analytics techniques and tools. We leveraged our experience from research in modeling Ebola spread (Shaw et al. Modeling Ebola Spread and Using HPCC/KEL System. In: Big Data Technologies and Applications 2016 (pp. 347-385). Springer, Cham) to successfully model Corona spread, we will obtain new results, and help in reducing the number of Corona patients. We closely collaborated with LexisNexis, which is a leading US data analytics company and a member of our NSF I/UCRC for Advanced Knowledge Enablement. The lack of a comprehensive view and informative analysis of the status of the pandemic can also cause panic and instability within society. Our work proposes the HPCC Systems Covid-19 tracker, which provides a multi-level view of the pandemic with the informative virus spreading indicators in a timely manner. The system embeds a classical epidemiological model known as SIR and spreading indicators based on causal model. The data solution of the tracker is built on top of the Big Data processing platform HPCC Systems, from ingesting and tracking of various data sources to fast delivery of the data to the public. The HPCC Systems Covid-19 tracker presents the Covid-19 data on a daily, weekly, and cumulative basis up to global-level and down to the county-level. It also provides statistical analysis for each level such as new cases per 100,000 population. The primary analysis such as Contagion Risk and Infection State is based on causal model with a seven-day sliding window. Our work has been released as a publicly available website to the world and attracted a great volume of traffic. The project is open-sourced and available on GitHub. The system was developed on the LexisNexis HPCC Systems, which is briefly described in the paper. 
    more » « less
  5. In this work, we show that for a nontrivial hypothesis class C, we can estimate the distance of a target function f to C (estimate the error rate of the best h∈C) using substantially fewer labeled examples than would be needed to actually {\em learn} a good h∈C. Specifically, we show that for the class C of unions of d intervals on the line, in the active learning setting in which we have access to a pool of unlabeled examples drawn from an arbitrary underlying distribution D, we can estimate the error rate of the best h∈C to an additive error ϵ with a number of label requests that is {\em independent of d} and depends only on ϵ. In particular, we make O((1/ϵ^6)log(1/ϵ)) label queries to an unlabeled pool of size O((d/ϵ^2)log(1/ϵ)). This task of estimating the distance of an unknown f to a given class C is called {\em tolerant testing} or {\em distance estimation} in the testing literature, usually studied in a membership query model and with respect to the uniform distribution. Our work extends that of Balcan et al. (2012) who solved the {\em non}-tolerant testing problem for this class (distinguishing the zero-error case from the case that the best hypothesis in the class has error greater than ϵ). We also consider the related problem of estimating the performance of a given learning algorithm A in this setting. That is, given a large pool of unlabeled examples drawn from distribution D, can we, from only a few label queries, estimate how well A would perform if the entire dataset were labeled and given as training data to A? We focus on k-Nearest Neighbor style algorithms, and also show how our results can be applied to the problem of hyperparameter tuning (selecting the best value of k for the given learning problem). 
    more » « less