by Aabir Abubaker Kar
1 min read

Categories

  • posts
  • projects
  • Projects

Tags

  • complex systems research necsi consulting strategy

Quick summary:

  • What I did:
    • data science consulting for a large manufacturing company
    • TAed a Summer School and a Winter School, helping professionals learn how to analyze data and code simulations using Python
    • co-authored a research paper on supply chain strategy
  • Things I learnt:
    • complex systems thinking
    • supply chain strategy
    • dimensionality reduction algorithms (PCA, t-SNE, MDS)
  • When was this?
    • September 2017 through July 2018

NECSI and Complex Systems Science

The New England Complex Systems Institute had been a ‘dream’ institute for my undergraduate thesis. I had been followed the work of Prof. Yaneer Bar-Yam for a few years. I couldn’t believe my eyes when I got my first email response from Alfredo Morales Guzman at the lab, someone I would go on to have a rewarding collaboration with.

Exposure to complex systems thinking was fundamentally paradigm-shifting for me as a 21-year old. The ideas of self-organization (order without hierarchy), emergence (global patterns from local interactions) and complexity (a measure of the amount of information stored in a phenomenon, or the rarity/improbability of a configuration) are transformative thinking tools.

What is a research consulting project?

While at NECSI, I worked on a research consulting project with an unusual client for a research institute. We were working with a manufacturing company that had $500 million in annual revenue - whose management were interested in better understanding their own business.

At the managerial level, decisions are often made on small scales of time and cost - seeking reliable, small improvements rather than dramatic, risky ones. This systematically blind-sides any organization to certain critical long-term views of itself - and can potentially obscure important information from the decision-making process.

Consider an example - Blockbuster was a company that once dominated the video-rental market in the US. In the late 90s, they had over 9000 stores. Netflix, spawned around 1997, attempting to change the market by offering mail-in DVD rentals. Blockbuster was slow to react to the changing market - they were reluctant to shift strategies from what had been working well for them. The rest is history. Blockbuster filed for bankruptcy in 2010, and as of 2020 is operating with a single store. Read more here.

The fall of Blockbuster is a strong example of how this lack of organizational self-awareness can promote the slow collapse of even the biggest companies. In rapidly changing times, that self-awareness becomes all the more important.

This project was a fine example of how a new set of eyes can often hold up the mirror to us, helping us improve in non-obvious ‘emergent’ ways. This company had 20 years of operations logs (messily) tucked away into SQL databases - they wanted to use the data better understand themselves, their growth, and their best strategies for the future.

What did I do?

At NECSI, my task was to use SQL and Python to extract and analyze these 100s of gigabytes of data. I worked extensively with dimensionality reduction algorithms on multiple types of data (geospatial, time-series, customer/product features), with our analysis culminating in a novel way to cluster and strategically target customers based on their purchasing behavior.

We also developed a new quantitative strategy for optimizing supply chain risk. We published this result in the journal Complexity under the title ‘Freight Time and Cost Optimization in Complex Logistics Networks’ - you can see the paper here.

While at NECSI I also had the chance to teach Python for Complex Systems to 2 cohorts of motivated Summer and Winter School students, as well as to present a poster at the International Conference on Complex Systems, 2018. I was lucky enough to hear some phenomenal speakers like Sandy Pentland, Nassim Nicholas Taleb, Stephen Wolfram, and many more - as well as to absorb the impressive intellectual culture of the Cambridge area.

My thanks to Prof. Yaneer Bar-Yam and Prof. Alfredo Morales-Guzman, as well as all my colleagues at NECSI for this amazing experience.