Our fourth and last use case has to do with smart cities. Now, a city is a geographically bounded space, and contains many different networks operating within the same spatial domain. What kind of networks? Let's see. It has transportation networks. Water and sewage networks, power transmission networks, your IP broadband networks. Some of these networks have multiple subtypes, for example transportation networks include the bus networks, the subway network, the surface street network, and the railway network, and so on. These networks form a physical infrastructure and therefore can be represented as graphs where each node has a geographic coordinate. But some of these networks can also be thought of as a commodity flow network. People flow through transportation networks. Sewage material flows through a sewage network and so on. For many of these networks, a city planner would like to make sure that they cover the entire city, that commute time is optimized, traffic congestions are well planned. To accomplish this, they would need to create what's called a network model. For example, urban planners develop city traffic model servicing. A traffic model will use both the geographic layout and connectivity of the network along with the flow parameters like the number of commuters getting on board at any station. Well, if we're planning to create a smart hub, we need to make sure that all the right things happen at the same place. People who come out of a metro station find nearby businesses. They find nearby facilities. Those facilities should have broadband network for people who are going to go on their mobile phones. The same places need to have a water supply network, and you have to plan it in such a way that all these networks who exist within a certain distance of each other. And all the facilities can be planned accordingly. Beyond normal operations, we also need to model what would happen if the network gets disrupted. What are the kinds of congestion or traffic that might disturb the network? Therefore these graphs are no longer just structures, but they represent things like congestions, things like people's behavior and materials behavior over the network. One should also compute energy use patterns for the busy parts of the network in order to figure out how the network structure or the network flow can be altered to enable energy optimal operations. As we saw in these four examples, what graphs are used for are kind of different but they all show different view points from which you can use graphs for analysis. So this course focuses on Graph Analytics. I would like to briefly recap what the term means. Analytics is the ability to discover meaningful patterns and interesting insights into data using mathematical properties of data. It covers a process of computing with mathematical properties and accessing the data itself efficiently. Further, it involves the ability to represent and work with domain knowledge as we saw in use case two with biology. Finally, analytics often involves statistical modeling techniques for drawing inferences and making predictions on data. With analytics, we should be able to achieve the goals shown here. Therefore, Graph Analytics is a special piece of analytics where the underlying data can be modeled as a set of graphs.