Studies on Risk Management as a Learning Analytics in Project-based Software Engineering Education
Year: 2018
Type: dissertation
Abstract: Similar to professional projects, project -based software engineering courses must consider the inherent risks and failures in software development processes.The fear of failure should not deter students from increasing their efforts and practising new ski lls.The risk of failure should be recognised by both students and teachers as learners who have failed are more likely to develop to their full potential as compared to learners who have never failed.However, assessments do not necessarily assist teachers' understanding of students' errors.Most assessors are required to measure the extent of students' accurate output by marking examination answers or practical assignments.Project-based learning gave birth to the situated learning concept, whereby learners construct their own learning experiences through context.However, at -risk learners cannot gain knowledge independently through project work.The present research addresses the questions 'How can risk management assess the specific factors of at-risk learners and how can individual learners be encouraged to gain skills and knowledge?'To investigate the research problem, this dissertation studies risk assessment at two levels.At the first level (the project level), it improves the fuzzy failure mode and effect analysis (fuzzy FMEA) to better understand risk assessment and improve the effectiveness of the risk priority number (RPN).This is achieved by (i) constructing an agent-based membership function and (ii) employing a voting fuzzy rule-based classification.The former technique provides an expert system that simulates the membership function of data-based systems on fuzzy inference systems, and the latter uses voting techniques for easy risk assessments by students.The main objectives of this experiment are to test the way of constructing the membership function and to test the process of classifying the proposed fuzzy rule.At the second level, the dissertation identifies individual and team level failures by (i) agentbased simulation of the input and possible data for subjectivity and risk analysis and (ii) designating assessments using a fuzzy signature (namely, a 'Tree Dependence Model').The agent-based model simulates the landscape of the characteristics of at -risk learners in each team.The teams comprise individual agents that engage in recognition -primed decision (RPD) and multi-level decision-making in team processes.The failure modelling in this research presents a diametrically opposed perspective to that of human-performance modelling.Meanwhile, the structure formed by the tree dependence model provides a history of each input datum, facilitating the examination of hierarchical positions and sequential ordering.In an actual classroom assessment, this model revealed that some assessing criteria act as weighing factors rather than judging factors.This study illustrates real-life contexts in which the plan is the predefined project plan, and the solution is the way of solving the problems of the current situation, in this case, the way of assessing the project risks, team failure and at-risk students.構成員の学習意欲のばらつきや脱落する構成員など,ソフトウェア開発演習 では教育ならではのリスクが存在する.PBLの担当教員はその学習効果をあげるために ,こうしたリスクを管理する必要がある.また,PBLでは学習者の自主性も求められる ことから,学習者自身がプロジェクトとしてのリスク管理を行うことも重要となってく る. ソフトウェア工学教育におけるPBLの有効性が指摘され,普及が進むにつれて,PBL の手法に関しても様々な研究が行われているが,PBLにおいて効果的あるいは効率的に リスク管理を行う手法については研究が進んでいないのが現状である. 本研究ではこの点に着目し,PBLによるソフトウェア開発演習のリスク,とりわけ, 脱落する学習者(演習についていけず学習目標の知識やスキルの習得ができなかった学 習者)の発生リスクに焦点をあて,そのための系統的なリスク管理手法の提案を目標と する. 「脱落する学習者のリスク要因を抽出し,脱落パターンを識別して,リスクを回 避するには,従来のリスク管理手法をどのように発展させればよいのか?」が本研究の 研究課題(research question)となる. この課題を解決するため,本研究では二つのアプローチをとった.一つはFuzzy FM EAの適用である.FMEA(Failure Mode and Effect Analysis)は設計の不具合や潜 在的な欠点を見出すための手法の一つであり,製品設計だけでなく,作業及び管理プロ セスを対象とした手法も提案されている.FMEAでは,リスクの度合いを計算し,それ に基づいた対策をとるために,故障が発生した場合の影響,発生頻度,発生時の検出可 能性の三つの評価指標に基づいたRPN(Risk Priority Number)を計算する.RPNの各 指標の評点付けは過去の開発実績等に基づいて予め定められた評価水準を利用するが, 適用分野によっては曖昧性を多く含み,各指標の評点付けが困難な場合がある.Fuzzy FMEAはファジー推論によってこの課題を解決する手法である.ソフトウェア開発, 特にPBLにおいては曖昧性を多く含む.そこで本研究ではFuzzy FMEAをベースとし たPBLにおけるリスク評価手法を考案し,その有効性を検討した.二つ目はマルチエー ジェントシミュレーションに基づいたリスク評価手法の構築である.本手法では,PBL におけるチーム開発をマルチエージェントモデルによってシミュレートし,チームレベ ルおよび構成員レベルでの学習者の脱落要因を特定し,リスク評価を行う.PBLにおけ るチームおよび構成員の振る舞いをシミュレートするために,プロジェクト構成員(す なわちPBLの学習者)をエージェントとしてモデル化する.ソフトウェア開発では様々
Author: Issarapong, Khuankrue
Language: English
Subfield: Education
Field: Social Sciences
Domain: Social Sciences
Keywords: risk management, learning analytics, at-risk students, fuzzy failure mode and effect analysis, risk priority number, membership functions, agent-based modeling, team failure, recognition-primed decision, multi-level decision making, software project risk assessment, student assessment, situated learning, failure models, expert systems, fuzzy inference system
SDG: Quality education
Open Access status: green
OpenAlex ID: W7145948789
