Any changes that would be made are only needed locally and will not be persisted. The data we send out is a static dataset that is produced by another group once a year representing historical data. Our users use this data as supplemental information to aid in building briefs. We were looking to improve the accuracy of their briefs by interpolating new values from the static dataset based on the user's input.
The original dataset is a table with something like 50 columns. Each row in the table represents a fixed polygon region on the globe. At each of these locations there are several measurements that are taken. Each measurement is split into a different column. So, for example, at X = 100 maybe Y = 32.5 for this particular location. Let's say there's 50 measurements meaning there's 50 columns with the headers "100", "200", "300"... "4900", "5000", etc. We symbolize the layer on the first column "100" by default with a class breaks renderer. The user may need pictures of the region where X = 600. In that case, we can just symbolize on the "600" column. However, in the case where the user wants to see the region at X = 560 we can use the data to interpolate the values at X = 560 to give the user a more accurate result. We thought of adding an extra empty column to our data just labeled "Interpolated" so that in the case where they want to see X = 560 we could set the class breaks to symbolize on "Interpolated" and then calculate the new values for each polygon without needing to edit any of the original data.
Our current software displays the closest measurement (If X = 560, we display the "600" column). Ideally, we want to be able to visually display the results from our interpolated values.
Again, the dataset is located on the user's system and will not be touched by any other user. We are only wanting to edit this fake "interpolated" column, and not the actual recorded data. The user knows the readings we would be providing would not be actual readings and are to be taken as approximations.
