Latest Contributions by JaySandhu

It is hard to provide an answer without context on what these costs mean and what they will be used for. E.g., what is a negative cost? What should a solver do with it? Note that in Network Analyst, costs, really impedance, is minimized when finding a path or to restrict how far a service area will extend to. Jay Sandhu ... View more

You can look into using the Tabulate Intersections tool: http://resources.arcgis.com/en/help/main/10.1/index.html#//000800000044000000 Jay Sandhu ... View more

You can use two different approaches: 1. You can assume that the census geography can be represented by its centroids. In that case you can run a Closest Facility analysis from the bus stops to see how many such census centroids are within your analysis distance. You can then sum up the racial profiles from the census points reached 2. You can assume that for each census geography polygon everyone lives in a uniform distribution. You can then generate service area polygons from your bus stops for the appropriate distance (generalized option, make sure hierarchy is off) and then use these resulting polygons to intersect with the census geography polygon. Then given the percentage of the polygon contained in the service area (use a ratio of new area to original area) you can summarize the racial mix. Jay Sandhu ... View more

You can map the candidate facility locations capacity to the capacity field when load the locations into a location-allocation layer. The "supply values" at the 80000 supply locations should be mapped as demand,  and then choose the Maximized capacitated Coverage as the type of analysis and solve for 10 facilities. Read more here: http://resources.arcgis.com/en/help/main/10.1/index.html#/Location_allocation_analysis/004700000050000000/ Jay Sandhu ... View more

There may have been some changes/bug fixes to the service area generalized method that is causing the changes you see. The polygon generation is very sensitive to slight changes in tolerances we use. Also, the generalized method does not guarantee that all edges that were not traversed will be excluded (i.e., it does not have holes for unreachable streets). You can try the detailed method and that will guarantee correctness in terms of streets that were traversed and those that were not traversed. Jay Sandhu ... View more

The service area polygons depend on how much the road network was traversed. So bring up the service area layer properties on the Line Generation tab, turn on Generate Lines and then turn on the Split lines at breaks. Solve it again. And now you can see how the lines are being traversed and what break values are present on them. Note: you may want to symbolize the lines on break values to see where they start/end. Examine the data to see why the final break value is not being reached. It is likely there is some restriction or disconnected edge that is not being traversed. Jay Sandhu ... View more

Service area polygons cover the street lines that can be reached within the specified distance or travel time break value. It's goal is not to cover blocks. What settings are you using? From the screen shot it looks you have set a very small trim distance that is causing the service area polygons to be trimmed to within a few meters of the outer set of street lines. Perhaps that is causing the "weird" shapes that you are seeing. Jay Sandhu ... View more

Is there hierarchy in the network and is that on by default when you do the service area. (I think not looking at the picture but just want to be sure). Can you turn on the Line generation and re-solve for both 10 and 10.1 and check if the lines are being generated in the the area the polygon was covering in 10? Jay Sandhu ... View more

Can you describe some more on the steps you are doing when you get this message? Is this a file geodatabase? Have you followed the NA tutorial and succeeded in creating any network dataset? For example: have this worked for you: http://resources.arcgis.com/en/help/main/10.1/index.html#/Exercise_1_Creating_a_network_dataset/00470000005t000000/ Jay Sandhu ... View more

If you have set all the demand points with the cutoff you do not to set the impedance cutoff on the layer property. By the direction of travel I mean that the distance from A to B may not be same as B to A due to one way and or turn and other restrictions. So when you set an overall cutoff of 20 minutes it is computing distances from all the facilities to all the demand points that can be reached from the facilities in 20 minutes. When you set a 20 minute cutoff on the demand points you are saying compute distances from demand to facilities that can be reached in 20 minutes. It may not be same. Anyway, it may not be significant issue but may change the solution slightly. As far as weighted cost, the goal is that when you add up all the weighted costs (demand weight multiplied by the travel time) from the chosen facilities to the allocated demand it will be the minimum. Yes a 0.5 weight will be less than .7 for same distance but what we are doing is locating facilities to allocate the weighted cost so that overall sum is minimum. If the demand is all 1 then we are minimizing overall travel cost. Jay Sandhu ... View more

When you do not put in the 20 minute cutoff and solve, are all the demand points set to provide their own 20 minute cutoff? Or is it a combination of 15 and 20 minutes? Generally there should be no difference in the solution. If there is, how different is it? Are all the 15 and 20 and 20 minute cutoff being honored? If there is any difference I can only attribute it to the travel direction. For example the travel time from the facilities to demand may not be same as demand to facilities due to one-way or other road restrictions. The demand cutoff of 15 to 20 means traveling from demand to facility should be that much. Where as overall impedance cutoff applies from the facilities to demand. So is the solution wrong between the two scenarios? Remember Location-Allocation is a heuristic and their can be many solutions that have the same overall objective values (we minimize sum of weighted cost). Jay Sandhu ... View more

If the points are exactly at the intersections/junctions, you can bring them as an additional feature source in your network dataset. And when you define the travel time attribute, you can bring in the impedance value from this traffic light (make sure it is in the same units as the travel times, e.g. minutes). This way any path going through that intersection will incur the delay from the traffic light. If you want to see how to model more feature classes and setting up the impedance attribute for them, take a look at the Paris network dataset tutorial. http://resources.arcgis.com/en/help/main/10.1/index.html#/Exercise_2_Creating_a_multimodal_network_dataset/00470000005v000000/ Jay Sandhu ... View more

What role do you want the traffic light to play in the network? Are the traffic light points located on edges or at intersections? Jay Sandhu ... View more

You will have to do this in three solves. One for each type of facility you want to find. So set up a CF layer with the incidents (block centroids) and then load in the first type of facilities and solve for 1 closest. Export the results. Then, unload/remove the facilities and load in the next set and solve for 1 closest and export. Repeat for the third set. Alternatively you can create the CF layer with the incidents and copy this layer twice. Now you have three layers. Load each set of demand points to them and solve for 1 closest and export/merge the results. Jay Sandhu ... View more