Agile Project Management Using Large Language

Models
Dhruva G Ishaan Shettigar Srikrshna Parthasarthy
Department of Computer Science & Department of Computer Science & Department of Computer Science &
Engineering Engineering Engineering
2024 5th International Conference on Innovative Trends in Information Technology (ICITIIT) | 979-8-3503-8681-3/24/$31.00 ©2024 IEEE | DOI: 10.1109/ICITIIT61487.2024.10580873

PES University PES University PES University

Bengaluru, India Bengaluru, India Bengaluru, India
pes1ug20cs131@pesu.pes.edu pes1ug20cs170@pesu.pes.edu pes1ug20cs692@pesu.pes.edu

Sapna V M
Department of Computer Science &
Engineering
PES University
Bengaluru, India
sapnavm@pes.edu

Abstract—Agile data-driven methodology encourages Subsequent innovations in software development

engineering leaders to measure their teams’ performance by gradually accommodated more realistic aspirations of
leveraging metrics for improving visibility, identifying potential consumers and the agility required to retain a viable
roadblocks, and increasing delivery velocity. The work competitive advantage. Most organizations today use a variant
presented here introduces a cutting-edge paradigm of a data- of the Agile philosophy [2], and it has continued to usher faster
driven approach to Agile project management, contributing to ways of taking software applications to the market.
the evolving research in project management methodologies. As
organizations strive to consolidate the competitive market, their The Agile management philosophy is founded on the
success is often measured by their agility and resilience. Such principles of increased customer collaboration, reduced
agility stems from the underlying management practices that an documentation, robust processes and the need to be constantly
organization embraces and is crucial for the effective planning adaptive [3]. While the philosophy has catalyzed software
and delivery of large-scale projects. While management innovation and disruption, the philosophy has failed to keep
philosophies have continued to evolve, organizations pace with a transformative age of software development and
specializing in software development have increasingly adopted has continued to stagnate in the process. In this context, this
Agile project management practices to keep up with a disruptive paper explores the use of emerging technologies to advance
landscape inundated with rapidly emerging technological existing Agile processes, to make them relevant to the current
advancements. As organizations have continued to expand
software development paradigm.
globally, the existing Agile practices have been laggard and
sluggish, unable to keep up with the demands of a modern II. AGILE PROJECT MANAGEMENT
market. In this context, the authors introduce an Agile project
management framework driven by Large Language Models A. Introduction
(LLMs) to facilitate efficient management of large projects. The Agile Manifesto was proposed in the early 2000s as a
contemporary alternative to a monolithic and documentation-
Keywords—Agile methodologies, project management,
driven software development approach, and it quickly became
generative AI, Large Language Models, Kanban
a forebearer for advancements in project management
I. INTRODUCTION strategies. Four values and twelve principles form the bedrock
of the Agile Manifesto, and they advocate a paradigm shift in
As enterprises become entrenched in delivering large- software development, prioritizing human-centric and
scale and cross-functional software projects to global clients, outcome-oriented approaches. The Agile philosophy is an
the risk of failure looms large with several potential umbrella of underlying management practices including the
roadblocks. While the technological landscape is dotted with widespread Scrum, Kanban and XP [4].
spectacular failures and mishits, there has been a concerted
effort to identify and evolve a consistent software
development lifecycle to meet the demands of a volatile
market. Initial development frameworks and practices (such
as the Waterfall model) relied extensively on a structured and
rigid downstream development lifecycle. Such models
mandated the need for collecting comprehensive requirements
at the beginning of the project and allowed little to no change
during software development and testing [1].
Such philosophy was rooted in the notion that frequent
changes and modifications would derail complex projects,
resulting in unreasonable budget overruns and missed
Fig. 1. Agile Software Development Lifecycle
deadlines. However, the evolution of the technological
landscape and the increasing agility demanded by a volatile The Agile philosophy outlines the need for delivering
market made such rigid models unsuitable and archaic due to software in frequent bursts, increased customer collaboration,
the quantum of specifications required and voluminous cross-functional synergy and flexibility in accommodating
documentation, and monolithic models quickly fizzled out. changing requirements as shown in Figure 1.

Authorized licensed use limited to: Centre de Recherches et des Technologies des Eaux. Downloaded on July 31,2024 at 07:15:17 UTC from IEEE Xplore. Restrictions apply.
 B. Agile Techniques
The Agile philosophy encompasses more than a dozen
techniques within the software development lifecycle,
allowing enterprises to tailor their implementation according
to their specific requirements and the nature of their work.
Some of these Agile practices have been elaborated below.
• Scrum – Scrum is a prominent agile project
management framework characterized by iterative
and incremental development methodologies.
Within the Scrum framework, interdisciplinary Fig. 2. A variant of the Kanban Board
teams engage in brief, time-bound sprints with the
objective of delivering a potentially deployable III. CONTEMPORARY CHALLENGES
product increment at the culmination of each Despite the remarkable leapfrogging in software
iteration. The Scrum teams are characterized by a development achieved with Agile practices, the existing
Product Owner and Scrum Master, who decide framework has continued to lag behind a vigorously evolving
Sprint Planning, Daily Scrum, Product Backlogs and technological landscape, threatening the philosophy with
Sprint Backlogs [5]. obsolescence in the absence of innovation. In this context, it
• XP – Key principles of XP (eXtreme Programming) becomes necessary to outline the contemporary challenges [9]
include pair programming, frequent code integration in the effective implementation of the Agile principles, paving
and testing, and test-driven development to ensure the way for a systematic enhancement and upgradation of the
rigorous development and testing practices. aforementioned practices.
Additionally, XP mandates collective code As organizations expand globally, the workforce has
ownership coupled with disciplined development increasingly become disparate and cross-national, enhancing
practices, making it efficient for large-scale projects the challenges in managing cycles of iterative software
with nebulous requirements resulting in frequent development across time zones and nationalities.
collaboration with customers. XP continues to thrive Organizations rely on sub-optimal Agile project management
as it is a lightweight and highly flexible development practices, often attributable to the lack of trained Agile
framework [6]. practitioners within the organization. Furthermore, the
• Feature-Driven Development – FDD involves a cultural shift required makes the adoption of Agile processes
structured five-step process, comprising the resource-consuming. Furthermore, the absence of data points
development of an overarching model, delineation of and feedback loops make project management inherently
a feature list, planning, designing, and building heuristic, making it exceedingly vulnerable to human errors,
features iteratively. Central to FDD is the concept of inaccurate project estimates and insufficient resource
features, denoting discrete, client-visible allocation. These compounds significantly in the downstream
functionalities prioritized based on business value. processes.
This methodology fosters a refinement-based The need for constant collaboration with customers or
development framework based on a client-centric clients may significantly impede the pace of development due
approach, ensuring efficient collaboration between to ambiguous requirements or infrequent interactions. Some
the development team and stakeholders [7]. of these impediments include lack of communication among
C. Kanban stakeholders, deferred replies and deferred communication,
unavailability of customers and frequently evolving
The genesis of Kanban can be traced back to the late 1940s requirements from different stakeholders [10].
when Taiichi Ohno, a key architect of the Toyota Production
System (TPS), developed Kanban as a visual scheduling and In an environment where consumer behaviour is dictated
production control system to optimize manufacturing by the responsiveness, agility and turnaround time of
processes. The term "Kanban" (看板) in Japanese translates to enterprises, it becomes essential to evolve contemporary
"visual card." The tasks are represented as tiles on a Kanban strategies and techniques to address the needs of a fiercely
board and this traces the task across the project lifecycle. As competitive market. This can be accelerated by leveraging
the task progresses through the development process, the tiles emerging technologies and management philosophies to
are accordingly moved across different phases on the Kanban mitigate the roadblocks prevalent in the current Agile
board (as shown in Figure 2). processes.
In the present milieu, the utilization of Kanban aims to IV. RELATED WORKS
enhance transparency throughout the software development
lifecycle, and it serves as an efficient methodology for A. Literature Survey
implementing Agile and DevOps principles [8]. The Bharat Choudhary and Shanu K Rakesh (2016) [11]
aforementioned framework emphasizes the need for attempt to coalesce the diverging techniques used in Agile
instantaneous communication of capacity and the unhindered project management, and elaborately discuss the
visibility of work progress. Visual representation using tiles methodologies adopted by respondents. Their research
on the Kanban board empowers cross-functional team outlines Scrum (evident in Figure 3) as the most popular Agile
members to glean real-time insights into the status of each framework. The authors further enumerate the different
work item, instilling a heightened level of transparency techniques, advantages and limitations of Adaptive Software
throughout the software development process. Development (ASD).

Authorized licensed use limited to: Centre de Recherches et des Technologies des Eaux. Downloaded on July 31,2024 at 07:15:17 UTC from IEEE Xplore. Restrictions apply.
 Sergey Bushuyev et al. (2022) [14] introduce a novel
metric for assessing the quality of project management
processes using entropy. Project management quality is
evaluated through the information project entropy,
representing the team's confidence in project outcomes. A
management quality scale and the entropy index are proposed
for continuous monitoring throughout the project life cycle, as
illustrated in Figure 5. Three pathways of project
informational entropy are explored, emphasizing the ideal
scenario representing a consistent decrease in entropy as the
project lifecycle progresses. This research proposes the use of
quantifiable metrics to evaluate project management
Fig. 3. 65% of the respondents used Scrum (N=125) practices.

A. Ahmed et al. (2010) [12] explore the impact of Agile

processes on the quality and productivity of software
development and deduce inferences based on the data
collected from enterprises in Pakistan. The authors considered
knowledge sharing, self-organizing teams, participation of
stakeholders, team size, formal training, responsiveness to
change etc. as metrics for analysing productivity and quality.
The authors propose a refined Agile lifecycle based on their
inferential research, incorporating a renewed focus on
refactoring and parallelized debugging. The research reports a Fig. 5. Information entropy of the project across its lifecycle
positive correlation between the implementation of Agile
processes and the productivity/quality obtained thereof. Shuo Zhang and Lan Jin (2020) [15] employ the Monte
Figure 4 illustrates the refined Agile development lifecycle. Carlo method to enhance the accuracy of software project
schedule management. The research attempts to address the
dismal success rates in the software industry by providing a
simulation-based solution to the uncertainty in project
durations. The method involves decomposing tasks, and
determining optimistic and pessimistic times. After 25,000
simulations, the authors achieve a reliable estimate of the
project's total duration, demonstrating the potential feasibility
and efficiency of Monte Carlo simulation in scheduling and
management. The probabilistic result is shown in Figure 6.

Fig. 4. Modified Agile lifecycle, with increased emphasis on Code

Refactoring

Nazish Saleem et al. (2021) [13] present a comprehensive

analysis on the evolutionary project management function
across different levels of operations. The authors explore the
evolutionary characteristics of project management in Fig. 6. Probabilstic schedule estimation using Monte Carlo method
software development, examining traditional, agile, and
global approaches. The research emphasizes the critical link B. Research Gaps
between effective project management and success, and Agile project management continues to evoke academic
recognizes the evolving nature of project lifecycles and cross- interest, but existing research remains vastly inadequate. With
cutting technological disruptions. The authors analyze the the rapid proliferation of Large Language Models (LLMs),
project management function across individual, project, team, their impact on Agile project management and scheduling
and company levels to infer the levers needed for informed remains to be explored. Furthermore, the Agile processes fail
decision-making. Their work underscores the growing to take into consideration relevant data points for continuous
complexity of project management, especially in the global refinement within the ambit of every organization, making it
context where unique challenges such as geographical, a largely enterprise-agnostic technique with little
temporal, socio-cultural, linguistic, and organizational customization. As customers demand near-instantaneous
distances need specialized project management practices., and responsiveness, the current frameworks fail to consider the
advocates for tailored approaches based on project need for intelligent task creation, task management and
requirements, beckoning further empirical research to workload management. Existing research can be attuned to
consolidate the existing research and address the nuances of enterprise data by coupling Agile processes with LLMs.
complex multilevel project management dynamics.

Authorized licensed use limited to: Centre de Recherches et des Technologies des Eaux. Downloaded on July 31,2024 at 07:15:17 UTC from IEEE Xplore. Restrictions apply.
 V. GENERATIVE AI & LARGE LANGUAGE MODELS VI. PROPOSED METHODOLGY
A. Introduction A. Architecture
The rapid emergence and disruptive popularity of General Figure 8 introduces an Agile project management
Artificial Intelligence (GAI) has ushered in a new paradigm of architecture using the PaLM large language model coupled
efficient computing [16]. While it remains early days for the with internal enterprise data.
advent of GAI, it has quickly superseded traditional
computational techniques, percolating to several industries.
Unlike traditional machine learning algorithms trained to
perform exceptionally well for specific tasks, GAI attempts to
use large corpus of data to model realistic environments, with
the ability of the model to replicate or accelerate human
behaviour to a remarkable extent in several tasks.
A Large Language Model (LLM) is a statistical deep
learning model that has undergone extensive training on a vast
multimodal dataset, and it possesses the capability to generate
text, generate multimedia content, assist in summarization and
perform several challenging natural language processing
(NLP) tasks. Such models are trained on several thousand
gigabytes of multimodal data, to facilitate the development of Fig. 8. Intelligent Agile project management using LLMs
a general artificial intelligence.
The premise of an omni-channel interface facilitates
While safety and accuracy of such LLMs need further customers and end users raising tickets, feature requests or
quantification and research, the models undergo continuous other requirements on an ad hoc basis. These are stored
fine-tuning and refinement based on a robust feedback loop, internally on a cloud-based MongoDB Atlas database for
enabling the model to stay relevant in the face of constant further processing and refinement.
disruption and emergence of newer data. Industries continue
to explore and leverage such large language models to A large language model (PaLM in this instance) is used to
automate several routine tasks. This paper attempts to explore augment the information obtained from the clients, end-users
the viability of using such large language models to accelerate or customers using a large corpus of training data. A single
Agile project management of large-scale software projects. query is augmented to include the following fields – task
name, description, teams capable of resolving the ticket raised
B. PaLM LLM and the priority. The information obtained is used to forecast
Pathways Language Model (PaLM) is a dense decoder- potential deadlines for the new ticket, enabling the project
only Transformer model trained on 540-billion parameters. manager or Scrum master to leverage internal enterprise data
PaLM uses the Pathways system [17] that adeptly orchestrates to estimate the potential endpoint, enhancing the capabilities
diverse parallel computations across thousands of accelerators of existing enterprise-agnostic project management processes.
through efficient scheduling, while coordinating seamless The proposed intelligent task augmentation and allocation
data transfers. Through extensive evaluations on numerous methodology delves into relevant enterprise-specific details
language understanding and generation tasks, PaLM such as historical project timelines and resource allocation
consistently demonstrates state-of-the-art few-shot patterns. Such holistic perspective empowers Agile project
performance, surpassing existing benchmarks by significant managers to take informed decisions on task scheduling,
margins. Figure 7 [18] shows the performance of PaLM on optimize resource utilization, and mitigate risks effectively
scaling, benchmarked on the “Beyond the Imitation Game” [19].
scale against other LLMs. Additionally, PaLM performs well
on multilingual NLP benchmarks, such as translation and text B. Synthetic Dataset
summarization. Due to the absence of a publicly available enterprise
dataset, we were compelled to create a synthetic dataset [20]
to simulate the necessary intelligent task allocation. Attributes
such as the number of personnel allocated and priority were
used to understand their correlation with the duration of the
ticket raised on the Kanban board. Figure 9 is a sample of the
dataset used.

Fig. 9. Synthetic dataset used for estimating the duration of a new ticket
Fig. 7. Scaling behavior of PaLM on a subset of 58 BIG-bench tasks.
The dataset was synthetically created by using generative
This paper introduces the use of PaLM to facilitate AI (i.e., ChatGPT), where a list of 250 tickets with their
automated Agile task creation and management and explores necessary attributes was generated along with their duration.
the performance of the model in this context. These were validated by a project management professional to
ensure that the synthetic dataset reflected the ground truth.

Authorized licensed use limited to: Centre de Recherches et des Technologies des Eaux. Downloaded on July 31,2024 at 07:15:17 UTC from IEEE Xplore. Restrictions apply.
 VII. RESULTS & DISCUSSION TABLE I. ACCURACY METRICS

The analysis of the results is two pronged, making it Sl. No. Regression Model MSE R2 Score
imperative to delve into the inferences of the results obtained. 1 Linear Regression 1.234 0.7433
These inferences have guided downstream design and
implementation choices. 2 Polynomial Regression 0.8492 0.8234

3 Ridge Regression 0.7340 0.8473

A. Regression Modelling
In a departure from existing Agile project management 4 Lasso Regression 2.449 0.4907
processes, the proposed methodology hinges on leveraging 5 Elastic-Net Regression 1.5829 0.6708
enterprise data to facilitate data driven task allocation. The
GAM (Generalized Additive
aforementioned synthetic dataset was subject to various 6
Model)
1.145 0.7618
regression modelling techniques to understand the correlation
7 CNN [1000 epochs] 0.9203 0.8086
of a set of attributes on the duration of a new ticket. The mean
square error (MSE) and R2 score (tabulated in Table 1) were
used to quantitatively determine the best performing The coefficient of determination (R2) quantifies the
regression models for further consideration. A consistent proportion of variance in the dependent variable explained by
train-test split of 80:20 was used with a random split across the model, while Mean Squared Error (MSE) measures the
the machine learning models during the analysis. Figures 10 – average squared difference between predicted and observed
14 visually illustrate the results obtained. values, serving as a measure of predictive accuracy in
regression analysis. Lower MSE values and higher R2 Scores
generally indicate better performance [21].
Ridge Regression (alternatively referred to as Tikhonov
regularization or L2 regularization) [22], is a linear regression
technique integrating a supplementary penalty term within the
cost function. This augmentation is proportionate to the square
of the coefficients' magnitudes, serving to mitigate the
influence of large coefficient values. Ridge Regression aims
to counteract overfitting by imposing constraints on the model
Fig. 10. Linear Regression and Polynomial Regression parameters, resulting in a more stable and generalized
predictive performance.
In the experiments conducted, the Ridge Regression
performed exceedingly well across both the quantitative
metrics measured. Therefore, the Ridge Regression model
was selected for further downstream implementation.
B. The Large Language Model
The PaLM LLM was considered for the research due to its
seamless integration into the programming environment via
publicly available APIs. The augmentation of an input ticket
Fig. 11. Ridge Regression and Lasso Regression (“Server is crashing on executing the program”) has been
tabulated in Table 2. Such augmentation by LLMs assist
project managers by infusing more context to incoming
tickets, and automates the routine hassle of raising tickets on
project management platforms.

TABLE II. SAMPLE TASK AUGMENTATION USING PALM

Task Field Output

Task Name Program crash on server
The server is crashing when the program is
Task Summary
executed.
Fig. 12. Elastic-Net Regression and GAM Regression Deadline 1

Priority High

Assigned to Backend (2), QA (1), Tech-Support (1)

Task types Bug-fix

Tags Java, Linux, Server, Crash

The above fields are auto-populated based on the incoming

request from the client, customer or end-user. Based on the
assignees suggested by the LLM, the duration is forecasted
Fig. 13. Regression using Convolutional Neural Networks (1000 Epochs) using the Ridge Regression model on the enterprise data.

Authorized licensed use limited to: Centre de Recherches et des Technologies des Eaux. Downloaded on July 31,2024 at 07:15:17 UTC from IEEE Xplore. Restrictions apply.
 VIII. CONCLUSION & FUTURE WORK [10] A. K. Nisyak, K. Rizkiyah and T. Raharjo, "Human Related Challenges
in Agile Software Development of Government Outsourcing Project,"
This paper attempts to identify the potential of LLMs for 2020 7th International Conference on Electrical Engineering,
automating Agile project management. This delves into a new Computer Sciences and Informatics (EECSI), Yogyakarta, Indonesia,
paradigm of data-driven Agile task scheduling and opens up 2020, pp. 222-229, doi: 10.23919/EECSI50503.2020.9251899.
new avenues of innovation in software management. [11] B. Choudhary and S. K. Rakesh, "An approach using agile method for
software development," 2016 International Conference on Innovation
Furthermore, the research introduces a novel coupling of large and Challenges in Cyber Security (ICICCS-INBUSH), Greater Noida,
language models with enterprise data. In this context, several India, 2016, pp. 155-158, doi: 10.1109/ICICCS.2016.7542304.
regression techniques for heuristic task scheduling have been [12] A. Ahmed, S. Ahmad, N. Ehsan, E. Mirza and S. Z. Sarwar, "Agile
elaborately discussed. software development: Impact on productivity and quality," 2010 IEEE
International Conference on Management of Innovation &
Future research on intelligent task-based scheduling and Technology, Singapore, 2010, pp. 287-291, doi:
reinforcement-based task management can usher in a new era 10.1109/ICMIT.2010.5492703.
of Agile project management. Furthermore, LLMs can be [13] N. Saleem, S. Mathrani and A. Sajjad, "Exploring the Evolutionary
effectively employed for analyzing screenshots uploaded by Characteristics of Project Management Approaches at Different Levels
customers for problem resolution. The next decade beacons a of Operations," 2021 IEEE Asia-Pacific Conference on Computer
Science and Data Engineering (CSDE), Brisbane, Australia, 2021, pp.
monumental disruption in traditional project management 1-7, doi: 10.1109/CSDE53843.2021.9718471.
processes.
[14] S. Bushuyev, S. Onyshchenko, A. Bondar, N. Bushuyeva and A.
Obronova, "Assessment and Monitoring Project Management Quality
REFERENCES Based on the Entropy Approach," 2022 International Conference on
[1] A. Ahmed, S. Ahmad, N. Ehsan, E. Mirza and S. Z. Sarwar, "Agile Smart Information Systems and Technologies (SIST), Nur-Sultan,
software development: Impact on productivity and quality," 2010 IEEE Kazakhstan, 2022, pp. 1-5, doi: 10.1109/SIST54437.2022.9945775.
International Conference on Management of Innovation & [15] S. Zhang and L. Jin, "Research on Software Project Schedule
Technology, Singapore, 2010, pp. 287-291, doi: Management Method based on Monte Carlo Simulation," 2020 IEEE
10.1109/ICMIT.2010.5492703. 5th Information Technology and Mechatronics Engineering
[2] B. Choudhary and S. K. Rakesh, "An approach using agile method for Conference (ITOEC), Chongqing, China, 2020, pp. 1605-1608, doi:
software development," 2016 International Conference on Innovation 10.1109/ITOEC49072.2020.9141570.
and Challenges in Cyber Security (ICICCS-INBUSH), Greater Noida, [16] M. Jovanović and M. Campbell, "Generative Artificial Intelligence:
India, 2016, pp. 155-158, doi: 10.1109/ICICCS.2016.7542304. Trends and Prospects," in Computer, vol. 55, no. 10, pp. 107-112, Oct.
[3] F. Paetsch, A. Eberlein and F. Maurer, "Requirements engineering and 2022, doi: 10.1109/MC.2022.3192720.
agile software development," WET ICE 2003. Proceedings. Twelfth [17] P. Barham, A. Chowdhery, J. Dean, S. Ghemawat, S. Hand, D. Hurt,
IEEE International Workshops on Enabling Technologies: M. Isard, H. Lim, R. Pang, S. Roy, B. Saeta, P. Schuh, R. Sepassi, L.
Infrastructure for Collaborative Enterprises, 2003., Linz, Austria, 2003, El Shafey, C. A. Thekkath, and Y. Wu, "Pathways: Asynchronous
pp. 308-313, doi: 10.1109/ENABL.2003.1231428. Distributed Dataflow for ML," 2022. Available:
[4] P. Sankhe, S. Mathur, T. B. Rehman and M. Dixit, "Review of an Agile https://arxiv.org/abs/2203.12533.
Software Development Methodology with SCRUM & Extreme [18] S. Narang and A. Chowdhery, “Pathways Language Model (PaLM):
Programming," 2022 IEEE International Conference on Current Scaling to 540 Billion Parameters for Breakthrough Performance,”
Development in Engineering and Technology (CCET), Bhopal, India, https://blog.research.google/2022/04/pathways-language-model-palm-
2022, pp. 1-6, doi: 10.1109/CCET56606.2022.10080640. scaling-to.html (accessed Jan. 4, 2024)
[5] A. Srivastava, S. Bhardwaj and S. Saraswat, "SCRUM model for agile [19] H. Wang, Z. Ren, Z. Yu, Y. Zhang, J. Liu and H. Cui, "CoupHM: Task
methodology," 2017 International Conference on Computing, Scheduling Using Gradient Based Optimization for Human-Machine
Communication and Automation (ICCCA), Greater Noida, India, 2017, Computing Systems," 2022 IEEE 28th International Conference on
pp. 864-869, doi: 10.1109/CCAA.2017.8229928. Parallel and Distributed Systems (ICPADS), Nanjing, China, 2023, pp.
[6] R. Juric, "Extreme programming and its development practices," ITI 859-866, doi: 10.1109/ICPADS56603.2022.00116.
2000. Proceedings of the 22nd International Conference on Information [20] M. Abufadda and K. Mansour, "A Survey of Synthetic Data Generation
Technology Interfaces (Cat. No.00EX411), Pula, Croatia, 2000, pp. 97- for Machine Learning," 2021 22nd International Arab Conference on
104. Information Technology (ACIT), Muscat, Oman, 2021, pp. 1-7, doi:
[7] S. Thakur and H. Singh, "FDRD: Feature driven reuse development 10.1109/ACIT53391.2021.9677302.
process model," 2014 IEEE International Conference on Advanced [21] D. Chicco, M. J. Warren, G. Jurman, "The coefficient of determination
Communications, Control and Computing Technologies, R-squared is more informative than SMAPE, MAE, MAPE, MSE and
Ramanathapuram, India, 2014, pp. 1593-1598, doi: RMSE in regression analysis evaluation," PeerJ Comput Sci., vol. 7,
10.1109/ICACCCT.2014.7019376. e623, Jul 5, 2021. DOI: 10.7717/peerj-cs.623. PMID: 34307865;
[8] M. O. Ahmad, J. Markkula and M. Oivo, "Kanban in software PMCID: PMC8279135.
development: A systematic literature review," 2013 39th Euromicro [22] A. M. A. Sai, K. L. Sai Eswar, K. S. Sai Harshith, P. Raghavendra, G.
Conference on Software Engineering and Advanced Applications, Y. Kiran and M. V., "Study of Lasso and Ridge Regression using
Santander, Spain, 2013, pp. 9-16, doi: 10.1109/SEAA.2013.28. ADMM," 2022 2nd International Conference on Intelligent
[9] W. R. Fitriani, P. Rahayu and D. I. Sensuse, "Challenges in agile Technologies (CONIT), Hubli, India, 2022, pp. 1-8, doi:
software development: A systematic literature review," 2016 10.1109/CONIT55038.2022.9847706.
International Conference on Advanced Computer Science and
Information Systems (ICACSIS), Malang, Indonesia, 2016, pp. 155-
164, doi: 10.1109/ICACSIS.2016.7872736.

Authorized licensed use limited to: Centre de Recherches et des Technologies des Eaux. Downloaded on July 31,2024 at 07:15:17 UTC from IEEE Xplore. Restrictions apply.