[MUSIC] In this lecture, you will study basics of Business GIS, which are Proximity Analysis and Accessibility Analysis. For Business GIS, the very first step is to figure out trade area, which is the area from which the store draws the most of its customers. And within which market penetration would be expected the most. For analyzing trade area, proximity analysis can be used. And accessibility analysis would be take one more step further to overcome limitations of proximity analysis. And present a more comprehensive insight of space associations of supply and demand. I've brought in the previous example presented in the very first lecture of this course, while we discussed special data science The question was, what is the revenue estimation of a new grocery store at the red arrow located in Evanston, IL? There are two approaches to answer the question. Causality approach and data mining approach. For causality approach, we may need to delineate trade area in the very first place. Let us assume that the blue line represents the trade area of a new grocery market. Actually, it's an example of proximity analysis, then we should collect additional information on potential customers, competing stores, field surveys, transportation, real estate price, and the new development and many others. Then we could come up with revenue estimation of the new store. Proximity analysis is by definition spatial analysis to determine distance relationship generally finding the closest pair between a selected feature and other features. There are two type of proximity analysis. One is to start from demand side the other one is from supply side. Find the closest stores from all the customer locations, that's the approach from demand side. On the other hand, delineating area coverage to which a given store has the minimum distance is the approach from supply side, confusing? Let's take a look at real-world axample, proximity to emergency rooms. The example is composed of district centers as customers locations, and locations of general hospitals with emergency rooms in Seoul Korea. Now, we can apply proximity analysis from demand side, which is customer's location. Basically, potential patient's locations and which is aggregated to the district centers. From each district center finding the closest general hospital with ER(Emergency Room) using simple geoprocessing tools. And the pairs are the result of the proximity analysis. Now you are looking at an example of proximity analysis for the entire Seoul. Pairing district center of black dot and the closest hospital with emergency room is the outcome of proximity analysis from demand side. One more thing should be noted that Euclidean distance was used for finding closest hospital from district center. However, there's a more realistic option using network distance, which can reflect the transportation condition. With respect to same dataset, Thiessen polygons are produced from locations of emergency rooms. Thiessen polygon is also known as Voronoi diagram which is partitioning of plane with input points into polygons, such that each polygon contains exactly one point. And any location in a given polygon is closer to its associated point than any other. So in the example, Thiessen polygon is the approximate area or area of influence from each general hospital which with emergency rooms. Which is an outcome, a proximity analysis from supply side. Now, you're looking at Thiessen polygons with respect to general hospitals with ER in Seoul, Korea as a result of proximity analysis. How do you like the results? Either analysis from supply side or from demand side. Generally speaking, proximity analysis is somewhat rudimentary and considers only distance or travelling time in defining trade area. However, we should consider other attraction factors and comprehensive analysis of both supply and demand together, for more practical business analysis. For that, the problems of proximity analysis could also be resolved by accessibility analysis. Accessibility is by definition, relative ease at which a service can be reached from given locations. Accessibility can be determined by the distribution of supply and demand and how they are related in space. So it is a classical example of spatial data science problem. Which is applicable to trade area analysis and resource distribution planning and so on. Supply-demand ratio is the simplest method for accessibility analysis. Which computes supply to demand ratio in an area, usually with respect to administrative district. However, the method have two serious shortcomings. First, it cannot review any variation within the areas. When the area is very large, it cannot reflect the reality. Second, it assumes no activity over the boundary. Demand is supported by supply only within areas which is far from realistic. You're looking at ER capacity of hospital to population ratio with respect to administrative district of Seoul, as an example of supply to demand ratio. The limitations of supply-demand ratio can be resolved by Floating Catchment Analysis(FCA). Instead of given areas, FCA defines a catchment area around each locations. And compute supply to demand ratio with respect to catchment area. The catchment area will float from one location to another. The boundary of catchment area can be delineated with the radius, travel distance or travel time. The figures illustrate examples of catchment area with radius and travel time, and two different outcome of R1 and R2. FCA method is much more improved over simple supply to demand ratio. However, it also has two logical problems. First, the distance between a supply and a demand within a catchment could be farther than a given threshold. Second, our supply center could be included in a multiple catchment. Which means that availability over supply centers, for example, ER capacity of hospital should be somehow discounted by the competition for it's service of surrounding demands. So for that 2-Step Floating Catchment Analysis was developed in order to overcome the problems of simple FCA. The solution requires two steps. First, applying supply to demand ratio in the catchment defined by each supplies center j which produce Rj. Second, find and sum all the js in the catchment of demand location i. Which produce AiF in the equation, which is accessibility of a demand location i Certainly a large value of AiF represents a better accessibility. For your understanding, I brought in an example. In the brown catchment of S2, two district centers, D1 and D2 of black dot are included. For ER capacity of S2 is divided by the population of the two districts, the result is R2. Likewise we can apply the same computation to S1 with green catchment and S2 with blue catchment. For each supply center, now we have supply to demand ratio, R1, R2, R3. For the second step, in a new a catchment in in green color around D1. Demand center, all the R values are added up and it is A1F, which is the accessibility of D1 from 2-step FCA. [SOUND] The accessibility can be also formulated by gravity model, which mimic gravitational interaction between two places. Like FCA, it also has two versions, single step or two steps. The single step also known as potential model is with respect to each demand at center i. The supplies from surrounding supply location j discounted by the inverse of the distance. Here, we assume that the beta is equals to 1. And the discounted supplies are all added up, and Ai is the accessibility of the demand center i. However, the potential model does not account for the demand side just like single step FCA method. So the potential model can be improved by considering competition for supplies among demand centers just 2-step FCA. The gravity-based index of accessibility AIG formulates the effect in terms of Vj, which is demand potential of supply location j. The demand from surrounding demand locations are discounted by the inverse of the distance, the discounted demands are all added up and you can get Vj. The Vj divided by discounted supply in previous model, now you have gravity based index of accessability, AIG. Again, here's an example. We have to start to compute competition for supplies. So with respect to S1, S2, S3, demand potential V can be computed. For example, S2 has two competing demand center, D1, D2 and the distance of 0.3 kilometer and 1.9 kilometer. So V2 can be computed by adding up the two discounted demands, likewise we compute V1 and V3. Then the demand potential divide each discounter supply, the terms added up. And now you can get a gravity-based index of accessibility of demand center D1 in the equation which is 0.002983. [SOUND] The two major approaches of accessibility are 2-step FCA and gravity-based index of accessibility. As you recognize, there are very similar to each other in computation. In reality, 2-step FCA ia a special case of the gravity-based method. The distance factors namely dij to minus beta, dkj to minus beta are only difference in the two equations, in the two models. So if the distance factor are removed from gravity model, they become identical. The two methods are applied to the test dataset of ER capacity of general hospitals and population of district in Seoul, Korea. In order to compute the accessibility of the ER for each district, For 2-step FCA five minutes travel distance were applied. And for gravity based approach, beta was set to 0.8. You are looking at accessibilty map produced by two methods. Then now your question should be which is better than the other? Actually it is a tough question, but generally 2-step FCA is preferred in most applications. Even though gravity-based index of accessibility is theoretically more sound than 2-step FCA. However first of all, gravity-based approach has a tendency to inflate accessibilty in poorly-accessed area and it generally requires large computation. Also, another problem is the distance friction factor, here the beta is the governing factor of gravity based approach but it is extremely difficult to determine the value in an appropriate manner. In this lecture, you studied proximity analysis and accessibility analysis. The one includes consumer-based analysis and store-based analysis, which are related to the delineation of trade area. The other includes supply to demand ratio, as the simplest version of accessibility analysis, 2-step FCA and gravity-based index of accessibility were introduced as more advanced and comprehensive method for accessibility analysis. This is the end of this lecture. It might be somewhat complicated, but this lecture will be successful if you could see the value and potential of spacial data science, particularly for business applications. Bye now, and hopefully see you all in the next lecture.