Module 3: Network Analysis
Lecture Summary
Introduction to Networks
The lecture begins with the introduction of networks. A network is defined as a collection of entities and relationships that can be represented by vertices and edges also called graphs. Network science is an emerging field that is based on the theory of graphs. The lecture mentions that understanding networks helps us understand disease propagation which after covid I am sure we can all appreciate. There are two major components of a network: Vertices and Edges.
As the lecture continues, there is a discussion that we generally have two types of networks: single mode and two mode (more than 1 kind of node). We are introduced to the concepts of directed (edges have direction) / undirected (edges do not have a direction) and unweighted/weighted (weight is the strength of the relationship) network types. (What Is the Difference Between an Undirected and a Directed Graph?, 2023) A subgraph of a network centered around a node is called an “ego” network. The lecture concludes with a discussion on how Network Visualizations are helpful to drive interesting insights from a network.
Introduction to Network Visualization
The lecture begins with the concept of information visualization which is introduced to help us explore patterns in complex networks. The lecture discusses different types of network layouts (force directed, geographical, circular, clustering, hierarchical). I think when most people think about networks they visualize a geographic layout. Finally, the lecture discusses how we can visualize our own ego network from the professional networking site, LinkedIn.
Network Properties
The lecture begins with a discussion of the structural properties of a network focusing on centrality. For degree centrality, an undirected network has 1 kind of degree centrality. For directed networks, we have in-degree (lead into a node) and out-degree centrality (lead out of a node). We are introduced to the concept of path which is any sequence of non-repeating nodes that connect two nodes. Shorter paths are generally more desirable with the exception of certain types of networks (disease propagation). The lecture continues to define different types of centrality to include betweenness centrality, closeness centrality and eigenvector centrality.
As the lecture continues we discuss how to identify the key players within the network which involves correctly interpreting the centrality measures within the network. We are introduced to 4 additional network measures reciprocity, density, clustering, and distances. We can breakdown network subgraphs into multiple connected components. Connected components can either be strong or weak. Finally, we discussed bridges (an edge when deleted increases number of connected components) and cliques (sub graph where all nodes are connected to each other).
Visualizing and Analyzing Networks Using Gephi
This is a short lecture to discuss the open source software tool called “Gephi”. This tool is useful for calculating network metrics and network visualization.
Using Networks for Analysis
The lecture begins with a discussion on how we can draw conclusions from network analysis. The lecture focuses both on visualization and mathematical analysis. Network metrics are used in conjunction with visualization. We discuss Google and Apple case studies around patent network applications. Using network analysis we can see that Apple works with a core of super inventors whereas Google works with many small teams of inventors working together. This can all be visualized through network analysis. The 2nd case study discussed is an ingredient network. One view is ingredients used in recipes. In another view, we look at ingredients used as substitutions. Throughout this case study, we can understand how networks can be used in prediction models, how network analysis is more than visualization and how we need to use a combination of metrics for an in-depth analysis of a network.
The objective of this Transporation Network Analysis is to reduce the total cost of transportation (which is essentially the same as the project that I will be managing). The case study derives insights from cost per ton, shipment data, number of shipments per store truck size, number of routes per truck. The case study uses graph theory to visualize the network where nodes are (stores) and each related pair of nodes is an edge.
The case study concludes by stated that by using a tool to visualize the network it allows the transportation/distribution team to work collaboratively with the continuous improvement engineer to find opportunities to optimize the Shanghai retailers network. (Saci, 2022)
References
Saci, S. (2022, Jan 22). towardsdatascience. Transportation Network Analysis with Graph Theory. https://towardsdatascience.com/transportation-network-analysis-with-graph-theory-55eceb7e4de4
What is Route Optimization? (n.d.). descartes. https://www.descartes.com/resources/knowledge-center/what-is-route-optimization#:~:text=The%20goal%20of%20route%20optimization,best%20meets%20the%20company%27s%20objectives.
What is the difference between an undirected and a directed Graph? (2023, July 31). geeksforgeeks. https://www.geeksforgeeks.org/what-is-the-difference-between-an-undirected-and-a-directed-graph/
Young, A. (2019, July 24). Functional Components of Best Transportation Management Systems (TMS). Blog. Retrieved November 28, 2024, from https://blog.intekfreight-logistics.com
Great write-up on the lectures, I really enjoyed your tie-ins to to how you're looking to take advantage of this material in your professional career as well as your associated case study, which gives a nice exploration into your future project! From what I've learned, I think Gephi is a good exploratory tool, but has some limitations depending on your business needs, things like large size-of datasets & real-time entries, not to mention perhaps other options might have some tailoring already towards network shortest-path optimization schemes. It's good that you're already thinking of ideas to apply these learnings practically!
ReplyDeleteThanks for reading my blog A.J! Do you have some practical experience with network visualization. I can see you pointed out some limitations of the tool. Can you expand on what it is that you mean by real-time entries? I am curious to know more about what you think is lacking.
DeleteHey Mike, I completely agree with your take on how valuable network analysis can be, especially when it comes to practical applications like route optimization or improving business operations. The way you tied the lectures into your upcoming Transportation Management System project really shows how this information isn’t just theoretical—it has real-world impact. The idea of using graph theory to optimize routes, visualize connections, and reduce costs makes so much sense. It’s exciting to think that tools like Gephi might offer enough insights to potentially cut down on licensing fees for additional TMS modules. That kind of efficiency and flexibility is exactly what makes this type of analysis so powerful.
ReplyDeleteWhat stood out most to me was how this knowledge is adaptable to so many fields. Whether it’s optimizing transportation, analyzing patent collaboration networks, or even predicting ingredient substitutions, the core principles of network analysis stay consistent. Visualization is a huge part of making these systems understandable, but combining it with metrics like centrality or clustering takes it to another level. It’s amazing how something as abstract as a network can be turned into actionable insights that directly impact performance and decision-making. I think this type of information is incredibly useful for not just understanding systems, but actively improving them. Your perspective really drives home how these concepts can be applied in practical, meaningful ways.
Thanks for reading my blog Christopher! This class has helped me to expand my view on networks outside of primarily IT or business operations. So I agree with you understanding that the core principles of network analysis stay consistent is a very useful paradigm to work and think from.
DeleteHi Mike, your blog entry offers a comprehensive and detailed summary of the lecture material, effectively covering key concepts like network properties, visualizations, and case studies. The connection between network analysis and the writer's future project on Transportation Management Systems (TMS) adds a practical and relatable dimension, showcasing how theoretical knowledge can be applied in real-world scenarios. The case study on Shanghai’s transportation network is particularly insightful and ties well to the broader topic of network optimization. I wonder how the tools like Gephi might integrate with existing systems like TMS to enhance decision-making... Overall, the blog provides a thorough and engaging exploration of the topic and very enjoyable to read.
ReplyDeleteThanks for reading my blog Andre! Since generally most TMS systems have some network optimization tool or package. I am wondering if the backbone of these applications or modules in the TMS are all built on the principles/algorithms that are outlined in the Gephi tool. Because to be honest, I do not understand within the TMS the framework or concepts that that the network optimization is engineered from. But, it something I am going to start to look into.
DeleteHi Mike! Thank you for your thorough and detailed overview of the module, and thanks for sharing the case study about traffic network analysis! Logistics is pretty far from my realm, but I find resource optimization to be fascinating and I can only imagine how useful network analysis could be for visualizing resource allocation needs across a network. Also, how cool that you get to use the concepts we're learning in class on a project you're working on in your career! One of the things that I've enjoyed the most about this class is learning about everyone's backgrounds and how the concepts we're learning about relate to roles and industries that are completely different from my own. Personally, I'm excited about the possibility of using Network Analysis to better understand learning objectives for the academic program I support, and how they map to our education program objectives. Cheers to surviving 587! :)
ReplyDeleteThanks for reading my blog Katherine! I'm am sure you will be able to many opportunities to use network analysis in academia! Good luck with the rest of the program!
DeleteHi Mike – Thanks for the wonderful summary of the lectures. I also greatly appreciate the discussion of transportation network analysis. I've always thought of traffic more in terms of the ebb and flow of daily rhythms and less in terms of transiting through nodes on a network. So, it hadn't directly occurred to me that one could use the tools of network analysis to address this. Not sure why it hadn't occurred to me, but here we are. Very cool idea and I'd be interested to see what could be done with it. All the best on your next course!
ReplyDelete