Lauren Rosenshein demonstrates how new spatial analytics and statistical tools in ArcGIS 10.1 can turn data into understanding.
00:01 One of the top trends in computing is analytics or turning data into understanding.
00:06 For some of us, this means spatial analytics - things like finding spatial clusters...
00:11 ...understanding trends, and predicting outcomes.
00:15 To tell us a little more about some exciting new spatial analytics and how we can share our workflows...
00:21 ...please welcome from the geoprocessing development team, Lauren Rosenshein.
00:30 Thanks, John. I'm really excited to show you guys three powerful new analytical tools...
00:37 ...that will help you solve problems and answer questions in new ways using ArcGIS 10.1.
00:43 What we're looking at here is gas price data by station in Southern California.
00:49 Just looking at the points on a map, it's pretty hard to tell if there's a pattern.
00:53 Maybe it looks like gas is a little more expensive in the Beverly Hills and Hollywood area in dark red.
00:58 But overall, it's hard to find a pattern. So we'll use a new tool called Grouping Analysis...
01:04 ...which will help us find and understand the spatial patterns that exist in our data.
01:10 We'll use a couple of different variables that get at different aspects of the price of gas.
01:15 So in this case, we have a price variable, a rank variable, and a trend variable.
01:21 What grouping analysis does is it creates distinct groups based on those gas price characteristics...
01:27 ...making sure that within groups, the features are as alike as they can be...
01:31 ...and between groups, the features are as different as they can be.
01:35 And, we have the option to constrain those groups spatially.
01:40 So immediately we can see there's definitely a spatial pattern here.
01:43 We can see our blue and our green groups along the cost, our red and our yellow groups in the inland area...
01:51 ...and another output of the grouping analysis tool is a PDF report. This report has lots of useful information.
01:57 But we'll focus on this summary report.
02:01 This shows us that our blue group along the coast is our most expensive gas area.
02:07 Our yellow group in the center of our study area is in the middle in terms of the price of gas...
02:13 ...and our green and our red groups are the cheapest areas for gas.
02:18 So now that we understand a little bit more about these spatial patterns, the next logical question is why.
02:25 What might help us understand or explain these spatial patterns?
02:30 Maybe it's something like the distance from major highways.
02:34 The closer you are to a major highway, the more expensive gas is. Or maybe it's something like income.
02:40 More affluent areas are areas where gas is more expensive.
02:45 If we take a closer look, that income variable is actually a little tricky to calculate.
02:51 What we don't want to do is oversimplify things by just assigning the income for that census tract...
02:56 ...to the gas station that's sitting inside of it.
03:00 And that's especially true when we have lots of gas stations that are on the borders of multiple census tracts.
03:06 A better way to model that would be to use a drive-time polygon around each of those gas stations.
03:12 So how do we get our income data from our census tract polygons into our drive-time polygons?
03:18 Well in 10.1, we can do this using a powerful new tool called Areal Interpolation.
03:24 Areal Interpolation lets us take our income data from our census tracts and create a continuous surface...
03:32 ...getting rid of those boundaries and making it really easy for us to get that income data into our drive-time polygons.
03:40 So now for this drive-time polygon, we have a predicted income using aerial interpolation of about $106,000 a year...
03:48 ...give or take a $3,000 standard error.
03:52 Now that we've calculated these variables, the next thing that we want to do is test these hypotheses...
03:59 ...using another new tool called Exploratory Regression.
04:03 I've created a model which runs Exploratory Regression once for each of the groups that we created using grouping analysis.
04:12 What Exploratory Regression lets us do is choose all of the variables that we think might be related to the price of gas.
04:19 In this case, these are socioeconomic and demographic variables, our distance variables...
04:25 ...and Exploratory Regression tests all the different combinations of those variables...
04:29 ...telling us which ones are consistently doing a good job of explaining the price of gas...
04:35 ...which combination of variables might help us explain the pattern.
04:39 I've created a report which outlines the results of those analyses.
04:43 What it's telling us is that our income variable that we thought was really important in the coastal area is very important.
04:50 But, in the inland area, it's all about distance from highway exits and distance from the interstates.
04:56 So two different geographies in our study area, two totally different sets of variables...
05:01 ...that help us understand the price of gas. So at this point, my analysis is done.
05:08 The next thing that I want to show you is how easy it is to share your analysis using the new geoprocessing package.
05:15 A very important first step for sharing any analysis is documenting our methodology.
05:21 So I've included our methodology in this report.
05:24 Now, in ArcGIS 10.1, it's really easy to share your analysis...
05:29 ...because we're introducing the new ability to share it as a geoprocessing package.
05:34 I can choose to upload my package right into ArcGIS Online.
05:38 I can include additional files like that PDF report that we just created which outlines our methodology.
05:44 I can include additional tools if that methodology is really a workflow of running multiple tools.
05:51 And when I share it, it's taking my input and my output data, my models including any nested models or scripts...
05:58 ...all those additional files, putting them into a geoprocessing package and sending them up to ArcGIS Online.
06:05 From ArcGIS Online, they can easily be discovered and used by other GIS professionals all over the world.
06:13 Yesterday, I sent a link to this package to my friend, Drew, who's out in Denver...
06:18 ...and wants to see what variables might help him explain the price of gas in his own area.
06:23 So let's check in with Drew and see how he's doing with his analysis. Hey, Drew.
06:28 Hey, Lauren. Thanks for sending over that geoprocessing package.
06:31 It has everything that I need to start analyzing gas prices here in Denver.
06:35 So I'm here on ArcGIS Online, and I'm going to open the geoprocessing package and when I do that...
06:41 ...it's going to start unpacking in ArcMap where I've already started looking at my Denver data.
06:46 So when I unpack the package, it's going to add some new data to the table of contents...
06:51 ...and it's going to add the tools and a report file down into the results window.
06:56 And I'll look at that report file to find the methodology and how to actually use some of the tools in this package.
07:02 If I want to use those tools on my own data, I simply open the tool and select my data and run the tool.
07:09 So this is going to find the groups in the data that will explain some of the spatial patterns for gas prices in Denver.
07:15 So here I have a couple interesting results.
07:18 You can see there's two groups that are in West Denver area and then two groups that are in the Denver metro area.
07:23 So the question that I want to answer then is why are there these different groups?
07:27 What makes those areas so different?
07:29 So I followed your methodology, and I'm going to actually use the Exploratory Regression model...
07:33 ...to find out what are those difference and why they have occurred.
07:37 So I've made a report that has some of this information.
07:40 I found that actually an elevation variable is one thing that's really important in those west groups...
07:45 ...as well as the distance to the interstate highways.
07:48 In those east groups, we see that it's the number of competitors nearby that are really determining the gas prices in Denver.
07:56 So that's it for our analysis. Thanks, again, for sharing that geoprocessing package.
08:00 I'm looking forward to using it more and sharing some of the findings with my coworkers.
08:05 Thanks, Drew. I think spatial analytics...yeah, that's cool, right?
08:15 I think spatial analytics are the heart and soul of GIS...
08:19 ...and being able to easily share our analysis will help all of us better understand our world.
© Esri 2013 http://www.esri.com