RAI Spatial Prediction Template

This template applies different interpolation methods to sparse spatial data and determines the algorithm with the best performance. Our algorithm produces a multivariate surface based on the multidimensional sparse points, where each dimension (or column) is considered independently. The measure used for comparisons is the R^2 and the best surface is represented in the Map Chart.

This is a great tool for quickly generating robust surfaces for multidimensional sparse points (like well locations, core samples, etc.). Further, it backfills missing values, so it’s possible to mix point data from different sources to generate a composite surface.


  • Load your Data
  • Create a Spatial Grid covering all the locations from your data
  • Creates random Training and Testing sets
  • Run five different Spatial Methods: Inverse distance weighting, kriging, Loess smoothing, mean, and splines
  • Performs two different data transformations: log and sqrt
  • Identifies the NAs and backfills predictions on these points
  • Outputs the R-Squared and the RMSE per method and transformation
  • Outputs are visualized using heat maps on the Map Chart and 3D Visualization tools


  • CreateGrid: Creates a Grid containing all the locations of the input data. The User can modify the input parameters Cell Size and Buffer
  • PredictSurface: Runs the Prediction Methods and select the best one. It treats missing values and outputs the Result Grid
  • PredictSurface_Custom:  Allows the User to select a method and transformation and explore the resulting predictions


  1. Replace the Points data table with your own data
  2. Run the Update Grid function to create the customized grid (input Cell size and Buffer)
  3. Mark the part of the Grid you want to use for prediction
  4. Select parameters for the idw and loess
  5. Select the number of points for the testing set
  6. Go to Edit >Data Function Properties > Select PredictSurface > Edit Parameters…. Then, choose the Points input parameter and update the columns using the following order: API, SURFLON, SURFLAT and other variables of interest.
  7. Run the Predict Surface function
  8. Explore the Histograms and the Heat Map of the different variables selected
  9. In the 3D Surface page select a transformation and method to run the Create User Surface function and visualize it in the 3D Graph

Leave a Reply

Your email address will not be published. Required fields are marked *

Download Details

RAI Spatial Prediction Template

20 sales


  • Your cart is empty.
Product Details
Release date: July 7, 2017
Last updated: July 7, 2017
Current version: 1.0
Software application type: Spotfire Template
File format: .dxp
File size: 5mb
Requirements: TIBCO Spotfire 7.6+
Posted In:  

Copyright 2017