Gephi Tutorial - How to use Gephi for Network Analysis

hello guys we're here to talk about network analysis today I have any of you guys heard of network analysis before I know one so the tool that we use for it is called Goffe it's a free tool and we'll just dive into first overview of what network analysis involves and then while I'm demoing the tools you guys can ask any questions as we go so first of all if networks all around us we have a lot of like simple data sets that we usually work with but we have a lot of you know very connected data in our world today and network analysis is are just a way of kind of trying to get a lot of value from that data figure data very easily and make it into ways that we can kind of unbiased we look at the data and shape the data and then see what trends be fine so like for emails you have like the people sending the emails and who they're sending it to with the other people and those would be like the connection between two people for like what you so you can look at like a email of a company and see who is sending emails or who between departments and where this good communication which is bad communication you can look at you know your social network and map out see where your friends are connected to each other you can look at research papers and who they co-author or who they cite and you kind of find the biggest players in certain areas of research very easily and like Google Maps is an example of it as well with mapping out how you get to some one place to another you have destinations and you have the streets or the airplane routes and anything between and then itself like Google Pagerank it relies upon network analysis just figure out who the major players are between websites so this is kind of an example of what it would look like we're talking about individual players like a would be the nodes like the little dots and then the lines between them are the connections so we look at the individual they're like the names variables we talk about those dots as being the nodes in the graph and then the edges are the connections between them and those edges can be like weighted so you can have like one street look the same but take certain lengths to traverse it and you have like direction so you can have like I cited someone my paper but they didn't cite me back and then we can go into Devon just quickly describe a couple metrics that are then use very useful centrality is a big one trying to get an idea of who were like the main players and network our degree centrality is what you would normally think of in terms of like it's just how many connections we'll have like how who has the most friends on Facebook and closest centrality looks at like you know what's the you know average distance from me to every other if it might listen if I'm one node what's the average all the distances between the notes other notes that I could reach so how long would take me on average to reach another node and that can be pretty useful like trying to figure out where to put your pizza place and queena centrality is also very interesting because it uh looks at something we call the least cost path which is say any look I need to nodes have a one least cost path which is like the most efficient way to get from that node to that node and you look at what other nodes fall if the Pennines to try to measure is how many how many least cost pads you fall upon as far as the network as a whole is concerned and the other thing we'll talk about is clustering so you can kind of easily identify different groups that are formed clustering coefficient is what you should talk about which can we get max one if it is the max one that means that every other node in that group references all of the other nodes in that group that's called a clique and we have a tool to plug these graphs into your JavaScript and manipulate in JavaScript demo that real quick called Sigma so you can see the webpage and kind of idea of what you can kind of do I have this very simply this is just something I plugged in a graph for in terms of what the actual code looks like you just have you grab a file a premade file like this or make your own file that has a lot of settings in here that you kind of just ignore and if you just put your exs file that you would get from Jesse into here you can then look at your own say Facebook Network very easily and hover over and see all different people among your friends and how they connect now I'm a dad quickly into what gessie actually looks like so we start off originally you need to put in like your data and for that looks like we can take a quick gander it's just like a text file I think this is Java technically but you have your note so you define and then you default to find the edges you can put like you know metadata on the nodes but and weights and stuff on the edges when once you get in here you just have a billion different options for weight and play around with like so I can see if I'm curious about the hell did you cook how many like how many people have how many connections in my group I can run quick data like this into the degree distribution in the degree is how many other connections a single node has then quickly like see that like you know there's couple people who have up to up to over 200 connections with it within my friends and a pond is very this is my old Facebook data I have a couple more friends now and more than just two three groups but I found this very curious originally when I found that a this is my high school friends this is a northwestern friends missus of Western fraternity friends so there's high correlation between these two groups still can be pointed out and then over in this section there's different ways you can organize your data so this is like just different algorithms that you can run that will then reposition your data trying to form find their way to make it look again at a useful format though efan really knows his stuff so I like using him you can come back a little bit and then there's a lot of options for it can either you can look at any eventual point click on it with this tool and see the properties you've laid on it and then you can also use a tool like this to color so the shortest distance between two points so let's choose blue and say we're right here and we wanted to get all the way over here it would theoretically color that blue oh there that that not worked so there so click somewhere in here and then the blue ones are the ones that are the shortest path between the two and there's also like options like coloring all the nodes that are nearest neighbor of some nodes so say we can use color you want to be able to easily tell as we say we know one node well and we want to be able to make sure we mark it it has their friends as important you can then click on any image rule no Dino would theoretically be coloring it right now not the smoothest running software ever but it has like many many many different tools built into it we have any general questions about this or network analysis as a whole oh yeah any network data so long as you have nodes and edges so there's a you would have to create it though there's also like a lot of like datasets out there already that you can get a hold of so I think in here you bring some examples of some basic social networks that they have someone's karate club and jazz musicians and stuff like that so there's a lot of network data out there aurilla that you can use plug right in here but I think the fun is with Sigma you can have it very easily to be like you know interactable on your website and you how you change in Jeffie affects it in terms of how it will appear like you can also have like this bar originally worked when I downloaded this when some reason just stopped working but this would have like more specific information about any individual know that you're clicking on hello actually it's working again sweet so what you can click on someone and see like they're like links and like some attributes that you've identified as well just do some wish numbers I'm wondering if that that might be the like a value on the edge like how you know cuz edges themselves can be weighted so it can be like what maybe I think this is like the misery all characters from the books might be like how often they like talk during the book some like that can also zoom over to see details better despite that alright thank you guys [Applause]

Channel: Fullstack Academy

Views: 84,379

Rating: 4.6370969 out of 5

Keywords: network analysis, network analysis tutorial, gephi, gephi tutorial, gephi examples, how to use gephi

Id: gcfAT8aMxuQ

Length: 10min 54sec (654 seconds)

Published: Thu Aug 17 2017