Monthly Archives: August 2018

Congratulations to two new Masters (Spring 2018)

Congratulations to the Spring 2018 Masters!

Congratulations to the new Spring 2018 Masters from my research lab!  This announcement is a few months late, nonetheless both of these individuals deserve to be congratulated on their achievements!

Alexander Rodriguez, Masters of Science, Data Science and Analytics

Spring 2018 Masters Thesis - Alexander RodriguezAlexander David Rodriguez completed the Masters of Data Science and Analytics.  I first met Alexander when he took my graduate course in Intelligent Data Analytics as a foreign-exchange undergraduate student from Peru.  He performed so incredible in my course that I immediately started talking with him about the possibility of continuing his graduate studies with me.  After returning to Peru and working one year in industry, he came back and joined the DSA program at OU.  He work for me as both a GTA and an GRA, funded in part by the NIST Center of Excellence on Risk-Based Community Resilience Planning. His MS thesis is entitled “Data-based Stochastic Network Mitigation”.  The abstract follows.

Data-based Stochastic Network Mitigation abstract: Current decision-support frameworks to assist mitigation planning do not include uncertainty and complexity of network failures, either one or both. To close this research gap, this thesis walks through a demonstration of the importance of including uncertainty in the decision analysis to later propose a novel methodology that employs simulation data that encapsulates both uncertainty and complexity of failures modeled by domain experts. Thus, this work is divided in two parts. The first part of this work examines how component importance measures fail to give the necessary intuition for mitigation planning in the light of uncertainty. The analysis is assisted by a novel component importance measure called probabilistic delta centrality that demonstrates how previously neglected stochastic considerations change decisions suggested. In the second part, a new paradigm for stochastic network mitigation is proposed. The approach leverages realizations from scenario event simulations to develop a probabilistic framework that supports constrained decision making. This scenario event simulation framework is capable of comprising component fragilities, correlation among random variables, and other physical aspects that affect component failure probabilities. On the top of that, a statistical learning model is built to enable a rapid estimation of post-disruption impact, which permits a metaheuristic to intelligently explore feasible discrete enhancements from mitigation strategies. The search for near-optimal solutions can be restricted by limited resources and potential political, social, and safety limitations. Two examples are presented to exhibit how this method provides detailed information for mitigation. The level of complexity embedded in search along with its detailed solutions are pioneering in network mitigation planning.

From this work Alexander and I have published one conference paper, and have two journal papers in progress.   Alexander will be pursuing his PhD in Computer Science at Virginia Tech starting Fall 2018.

Yanbin Chang, Masters of Science, Industrial and Systems Engineering

Yanbin completed his Masters of  Industrial and Systems Engineering in the Spring as well.  His MS thesis is entitled “Heuristic approach to network recovery”

Abstract: This study addresses optimization modeling for recovery of a transportation system after a major disaster. In particular, a novel metric based on the shape of the recovery curve is introduced as the objective to minimize.  This metric is computed as the distance from the pre-disaster system performance at a time immediately before disruption to the two-dimensional location of the centroid point of the area beneath the recovery curve. The recovery trajectories derived from optimization models with this new metric are considered along with two other recovery goals from literature, i.e., minimizing the total recovery time and minimizing the skew of the recovery trajectory. A genetic algorithm is implemented to search for optimal restoration schedules under each objective and empirical analysis is used to evaluate the corresponding quality of the solutions. Additionally, a particle swarm optimization algorithm is employed as an alternative metaheuristic and the quality of the recovery schedules, as well as the observed computational efficiency is analyzed.

Yanbin is currently preparing this thesis work for submission as a journal article.  He will begin his PhD in Industrial Engineering at Clemson this Fall as well.

BONUS Material

Brad Osborn, Bachelors of Science, Industrial and Systems Engineering

Brad Osborn, completed his undergraduate in Industrial and Systems Engineering.  While Brad was not officially a part of the OU Analytics Lab, he worked with AT&T as an intern and used some of the skills he mastered in my class ISE 4113 Spreadsheet-based Decision Support Systems to wow his superiors.  They offered him a full-time job and relocated him to Seattle, WA.  However, his relocation is also bittersweet — Brad was a key player on my soccer team “Total Chaos” in the Norman area adult soccer league.  Regardless, I wish you great success at AT&T!

Congratulations Vignesh!

Congratulations to Vigneshwaran Dharmarajan on his new job!

Congratulations Vignesh on your new job!

Congratulations Vignesh!  Vigneshwaran just received a job offer as a data scientist and has relocated to Kansas.  Vignesh recently completed his Master’s of Science in Data Science and Analytics in the Gallogly College of Engineering at the University of Oklahoma and he also worked for me as a TA in the Fall of 2017 for the ISE/DSA 5103 Intelligent Data Analytics course.

Vignesh came by to see me the other day to let me know about his recent success and then sent me a note describing a little bit about the interview process and how his education in the DSA program helped him land the job.

I would like to thank you so much for offering the Intelligent Data Analytics (IDA) course. The knowledge that I acquired through the course work helped me get job offer as Data Scientist. As part of the interview process, I got a case study where I was given a real world data set to analyze, predict and to present the results that decisions can be made from. From the course work, I have learnt techniques like exploratory data analysis, feature engineering, train model, validate the model & future prediction. With this I can be able to perform all the requirements given in the case study without any difficulty and used all types of visual aids learnt in the course work to present the data more palatable or comprehensible. This helped me moved to the next round and in the final round the discussion started with the learning & the projects did in the IDA course along with other experiences & projects. I know, the knowledge and techniques I learnt  in the course is a significant factor for getting this job offer and will definitely help me to apply this in the industry to improve the business process. Thank you once again!

—  Vigneshwaran Dharmarajan

I have heard similar stories lately regarding what companies seem to care about in hiring new data scientists.  They regularly require candidates to address case studies and data sets as a first step in the interview process.  I am glad that the DSA program helped prepare Vignesh for this.

Vignesh also supplied me with the job description (excerpted below with highlights added).

Job Description

Provide accurate and timely data support, analysis and maintenance using statistical math and algorithms on a variety of reports, charts, models and projects that have a direct impact on all aspects of the organization.

Build Efficiency

  • Work closely with the Lead Data Scientist regarding efficiency, analysis, quality goals, and market and product trends
  • Contribute using SQL and R language to the design, development, and maintenance of ongoing metrics, reports, analyses, dashboards, etc. to drive key business decisions.
  • Support cross-functional teams on the day-to-day delivery of projects and initiatives
  • Find, clean and integrate data into usable and helpful information
  • Identify and utilize data analysis and measurements effectively to identify performance trends, shortfalls
  • Identify campaigns, strategies, and processes that drive highest results
  • Turn data into insight: segment, cluster, model and mine to better understand behaviors and trends
  • Analyze why things happened and predict what could happen in the future based on data

Maintain Operational Standards

  • Identify gaps in current data systems, drive and contribute to development of systems to bridge gap
  • Plan and execute ongoing strategies including maximization of technologies and processes
  • Take initiative to make things better and provide solutions for challenges

Developing People

  • Model Customer Service principles and practices and foster an environment where those principles guide accomplishing goals and interacting with internal and external Customers
  • Anticipate operational needs based on volume/expansion and make recommendations
  • Drive change produce clear, understandable visualizations and reports to share with Senior Management. Partner with management to design tests and implement your finding

As a quick note: in the ISE/DSA 5103 course you will learn and use R, learn how to clean and deal with messy data, work very hard on turning data into insight, use visualizations to explore and explain data, and learn an array of techniques for predictive modeling, among other things.

I will post more student’s stories on data science interviews and jobs in the future.   And if you are a former student who wants to be highlighted in the blog, just send me your information and picture and I am happy to boast about your success!