Darshan Nayak

Full Stack Developer

Research Works

These are the research papers I have written in fulfillment of various academic courses at the University of Regina.


Time Series Forecasting using Multiple Granularities and Multiple Forecast Models (M.Sc. Project Report)

Abstract: We evaluate the effectiveness of combining multigranular time series forecasts from multiple forecast models to improve time series forecasting accuracy. Time series forecasting is widely used in scientific and commercial activity since it provides a way of predicting the future by looking at the past. We start by generating multiple time series datasets of varying granularities from the original time series using the means, and generate a forecast for each dataset using two families of time series forecast models. Next, we assess the effectiveness of a number of techniques for forecasting. First, we start with generating forecasts using base granularity without any combinations. We use these forecasts as benchmark results to compare other results to. The first combination method that we use is the inter-model unigranular forecast combination using the Weighted Majority Average-Continuous Version (WMA-CV) algorithm, where we combine forecasts of original granularity from different families of models using WMA-CV to assess their effectiveness at maximizing forecast accuracy. We evaluate the strategy of combining multigranular forecasts using a linear combination of the means. We also propose to combine multigranular forecasts using the WMA-CV algorithm to models from the same family and models from different families.

Our results show that combining multigranular forecasts using the means or the WMA-CV algorithm is not effective at improving forecast accuracy, with the former producing worse results. However, combining forecasts of the original granularity from different families of models using the WMA-CV algorithm yields more accurate forecasts than forecasts without combinations.

Supervisor: Dr. Howard J. Hamilton

Download the Project Report submitted to the Defense Committee here .

Systematic Literature Review: Deciding between Agile and Plan-based Processes

Abstract: The software development community is reporting a huge spike in the demands of software. Amid a rapidly changing business culture, clients require rapid deployment of software to keep up with the competition, which has opened new opportunities for evolution of the development process with the recent introduction of agile practices. However, the benefits of choosing agile over traditional practices are still a point of debate in the community. A number of factors affect this critical choice and I attempt to address this matter in this report. I present a systematic literature review aimed at revealing the current practices of software development and how they affect the development teams and the projects. I also present my hypothesis based on the findings of the review.

Course: Advanced Software Process (ENSE 874)

Download the Research Paper here .

Process Improvement in Scrum using Mixed Research Methods

Abstract: Since the introduction of agile methodologies, the software development community has witnessed a radical change in the way software is developed. Development of code is now given more emphasis than documentation of the system. Scrum is an agile software development framework which has gained popularity and is used widely. A majority of teams which migrate to Scrum from traditional development have reported that they were able to reap better benefits from Scrum. It would be interesting to further enhance Scrum’s ability to gain more benefit. In this report, I put forward a number of ways people have modified Scrum to suit their interests. I also present a method for improving Scrum and present an experimental design to test the effectiveness of this method. The proposed design uses mixed methods of research to monitor this improvement.

Course: Advanced Software Process (ENSE 874)

Download the Research Paper here .

Significance of Animated Transitions in Mobile User Interfaces

Abstract: Animations or animated transitions are an important part of user interface (UI) design in mobile devices. They help in making the interface attractive, easy to use and contribute towards a better user experience. Animations also help in guiding user’s attention in response to some changes occurring in the interface. In the time-line of mobile devices, manufacturers have always used animations. Although animations were used sparingly in the initial years, we can observe an exponential rise in use of animation over the recent years, and users who have used both, the primitive mobile devices as well as the current high-end devices, will gladly agree to the fact that animations have ultimately led to a serene experience with their mobile devices. In this position paper, I will discuss ideas put forward by 3 research papers about Animated Transitions, their importance in mobile devices and some guidelines about how their power can be harnessed to improve user experience. The aim of this work is to cast light on the various ways in which animations have been used to render a captivating experience to the users in the world of mobile computing.

Course: Mobile Computing (CS 855)

Download the Research Paper here .