Towards Learning Analytics on US Navy Training Data
Abstract
While there have been efforts to deploy predictive and learning analytics in the military, there is still considerable room to enhan""ce the degree to which learning analytics supports training and human development in military contexts. In the proposed project, we"" hope to conduct research that can demonstrate the feasibility of learning analytics for the US Navy~s training efforts, as well as"" providing a roadmap for future efforts along these lines. Specifically, we will demonstrate that it is feasible to identify trainee""s at-risk for poorer outcomes, and trainees who are likely to excel, and that it is feasible to determine which trainees are likely"" to be benefitted by specific learning experiences, based on data currently available to the US Navy. Within this initial prelimin""ary project, we will attempt to study two research questions:RQ1. Can we infer which trainees are at-risk for poorer post-training"" performance, based on actionable features of their behavior? Relatedly, can we infer which trainees are likely to excel, post-train""ing, based on actionable features of their behavior?RQ2. Can we infer which training experiences are most likely to benefit indivi""dual trainees?More specifically, we will develop initial models that attempt to answer each of these two research questions, for a"" subset of possible trainees, indicators of success, and potential predictors. We will do this in the context of existing US Navy an""d military databases: exploring these databases~ potential, distilling meaningful features and predictors from the data sets, and de""veloping a roadmap for further work to research, develop, and deploy learning analytics to benefit trainees in the US Navy. This p""roject will have four phases: I) Data Exploration, II) Data Distillation and Feature Engineering, III) Creating Analytics Models, an""d IV) Creating Roadmap.In the first phase, Data Exploration, we will obtain data from relevant naval and military databases that c""an be used to understand trainee background, training experiences, and training outcomes, and we will explore this data to understan""d what variables are feasible for use in Phase II.In the second phase, Data Distillation and Feature Engineering, we will distill"" a representative set of each of four categories of trainee variables for use in Phase III, conducting an iterative process of varia""ble selection, prototype variable creation, and then replicating the prototype feature creation at scale using Python within an Amaz""on Web Services setup.In the third phase, Creating Analytics Models, we will conduct analysis on the variables distilled in the se""cond phase to preliminarily assess the feasibility of answering each of the two project research questions. For RQ1, we will build c""lassification prediction models that can infer whether a trainee is likely to succeed or not, and whether they are likely to have ex""cellent outcomes or not. For RQ2, we will distill a list of findings as to which training experiences are particularly likely to be"" effective for which trainees.In the fourth phase, Creating Roadmap, we will generate a roadmap for future research and developmen"t on how to leverage the US Navy~s wealth of educational data in order to achieve the goals listed above. This roadmap will include" multiple potential plans for data enhancement, as well as for future analysis, modeling, and intervention that can be conducted eit""her by our team or by other teams.At the conclusion of the project, the Naval Research Office will have a demonstration of the fea""sibility of using learning analytics on existing databases to enhance trainee outcomes, as well as information on possible paths for"ward for leveraging the excellent resources already available to enhance capacity.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jul 07, 2017
- Source ID
- N000141712662
Entities
People
- Ryan Baker
Organizations
- Office of Naval Research
- United States Navy
- University of Pennsylvania