Early prediction of student difficulty during long-duration learning activities allows a tutoring system to intervene by providing needed support, such as a hint, or by alerting an instructor. To be e effective, these predictions must come early and be highly accurate, but such predictions are difficult for open-ended programming problems. In this work, Recent Temporal Patterns (RTPs) are used in conjunction with Support Vector Machine and Logistic Regression to build robust yet interpretable models for early predictions. We performed two tasks: to predict student success and difficulty during one, open-ended novice programming task of drawing a square-shaped spiral. We compared RTP against several machine learning models ranging from the classic to the more recent deep learning models such as Long Short Term Memory to predict whether students would be able to complete the programming task. Our results show that RTP-based models outperformed all others, and could successfully classify students after just one minute of a 20- minute exercise (students can spend more than 1 hour on it). To determine when a system might intervene to prevent incompleteness or eventual dropout, we applied RTP at regular intervals to predict whether a student would make progress within the next fi ve minutes, reflecting that they may be having difficulty. RTP successfully classifi ed these students needing interventions over 85% of the time, with increased accuracy using data-driven program features. These results contribute signi ficantly to the potential to build a fully data-driven tutoring system for novice programming.
more »
« less
This content will become publicly available on July 1, 2024
Learning When to Defer to Humans for Short Answer Grading
To assess student knowledge, educators face a tradeoff between open-ended versus fixed response questions. Open-ended questions are easier to formulate, and provide greater insight into student learning,vbut are burdensome. Machine learning methods that could reduce the assessment burden also have a cost, given that large datasets of reliably assessed examples (labeled data) are required for training and testing. We address the human costs of assessment and data labeling using selective prediction, where the output of a machine learned model is used when the model makes a confident decision, but otherwise the model
defers to a human decision-maker. The goal is to defer less often while maintaining human assessment quality on the total output. We refer to the deferral criteria as a deferral policy, and we show it is possible to learn when to defer. We first trained an autograder on a combination of historical data and a small amount of newly labeled data, achieving moderate performance. We then used the autograder output as input to a logistic regression to learn when to defer. The learned logistic regression equation constitutes a deferral policy. Tests of the selective prediction method on a held out test set showed that human-level assessment quality can be
achieved with a major reduction of human effort.
more »
« less
- Award ID(s):
- 2010483
- NSF-PAR ID:
- 10418179
- Date Published:
- Journal Name:
- International Conference on Artificial Intelligence in Education
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Computer-aided design (CAD) programs are essential to engineering as they allow for better designs through low-cost iterations. While CAD programs are typically taught to undergraduate students as a job skill, such software can also help students learn engineering concepts. A current limitation of CAD programs (even those that are specifically designed for educational purposes) is that they are not capable of providing automated real-time help to students. To encourage CAD programs to build in assistance to students, we used data generated from students using a free, open-source CAD software called Aladdin to demonstrate how student data combined with machine learning techniques can predict how well a particular student will perform in a design task. We challenged students to design a house that consumed zero net energy as part of an introductory engineering technology undergraduate course. Using data from 128 students, along with the scikit-learn Python machine learning library, we tested our models using both total counts of design actions and sequences of design actions as inputs. We found that our models using early design sequence actions are particularly valuable for prediction. Our logistic regression model achieved a >60% chance of predicting if a student would succeed in designing a zero net energy house. Our results suggest that it would be feasible for Aladdin to provide useful feedback to students when they are approximately halfway through their design. Further improvements to these models could lead to earlier predictions and thus provide students feedback sooner to enhance their learning.more » « less
-
Advancements in online learning platforms have revolutionized education in multiple different ways, transforming the learning experiences and instructional practices. The development of natural language processing and machine learning methods have helped understand and process student languages, comprehend their learning state, and build automated supports for teachers. With this, there has been a growing body of research in developing automated methods to assess students’ work both in mathematical and nonmathematical domains. These automated methods address questions of two categories; closed-ended (with limited correct answers) and open-ended (are often subjective and have multiple correct answers), where open-ended questions are mostly used by teachers to learn about their student’s understanding of a particular concept. Manually assessing and providing feedback to these open-ended questions is often arduous and time-consuming for teachers. For this reason, there have been several works to understand student responses to these open-ended questions to automate the assessment and provide constructive feedback to students. In this research, we seek to improve such a prior method for assessment and feedback suggestions for student open-ended works in mathematics. For this, we present an error analysis of the prior method ”SBERT-Canberra” for auto-scoring, explore various factors that contribute to the error of the method, and propose solutions to improve upon the method by addressing these error factors. We further intend to expand this approach by improving feedback suggestions for teachers to give to their students’ open-ended work.more » « less
-
null (Ed.)Abstract: Modeling student learning processes is highly complex since it is influenced by many factors such as motivation and learning habits. The high volume of features and tools provided by computer-based learning environments confounds the task of tracking student knowledge even further. Deep Learning models such as Long-Short Term Memory (LSTMs) and classic Markovian models such as Bayesian Knowledge Tracing (BKT) have been successfully applied for student modeling. However, much of this prior work is designed to handle sequences of events with discrete timesteps, rather than considering the continuous aspect of time. Given that time elapsed between successive elements in a student’s trajectory can vary from seconds to days, we applied a Timeaware LSTM (T-LSTM) to model the dynamics of student knowledge state in continuous time. We investigate the effectiveness of T-LSTM on two domains with very different characteristics. One involves an open-ended programming environment where students can self-pace their progress and T-LSTM is compared against LSTM, Recent Temporal Pattern Mining, and the classic Logistic Regression (LR) on the early prediction of student success; the other involves a classic tutor-driven intelligent tutoring system where the tutor scaffolds the student learning step by step and T-LSTM is compared with LSTM, LR, and BKT on the early prediction of student learning gains. Our results show that TLSTM significantly outperforms the other methods on the self-paced, open-ended programming environment; while on the tutor-driven ITS, it ties with LSTM and outperforms both LR and BKT. In other words, while time-irregularity exists in both datasets, T-LSTM works significantly better than other student models when the pace is driven by students. On the other hand, when such irregularity results from the tutor, T-LSTM was not superior to other models but its performance was not hurt either.more » « less
-
The US Drought Monitor (USDM) is a hallmark in real time drought monitoring and assessment as it was developed by multiple agencies to provide an accurate and timely assessment of drought conditions in the US on a weekly basis. The map is built based on multiple physical indicators as well as reported observations from local contributors before human analysts combine the information and produce the drought map using their best judgement. Since human subjectivity is included in the production of the USDM maps, it is not an entirely clear quantitative procedure for other entities to reproduce the maps. In this study, we developed a framework to automatically generate the maps through a machine learning approach by predicting the drought categories across the domain of study. A persistence model served as the baseline model for comparison in the framework. Three machine learning algorithms, logistic regression, random forests, and support vector machines, with four different groups of input data, which formed an overall of 12 different configurations, were used for the prediction of drought categories. Finally, all the configurations were evaluated against the baseline model to select the best performing option. The results showed that our proposed framework could reproduce the drought maps to a near-perfect level with the support vector machines algorithm and the group 4 data. The rest of the findings of this study can be highlighted as: 1) employing the past week drought data as a predictor in the models played an important role in achieving high prediction scores, 2) the nonlinear models, random forest, and support vector machines had a better overall performance compared to the logistic regression models, and 3) with borrowing the neighboring grid cells information, we could compensate the lack of training data in the grid cells with insufficient historical USDM data particularly for extreme and exceptional drought conditions.more » « less
