User Guide for W-P-MLR models

If you are using the MATLAB toolkit as a function:

Step 1: Download the MATLAB tool-kit for execution of Wavelet based multi-resolution Multiple Linear Regression (W-P-MLR) models from the "Files" section in the navigation panel on the left side of the webpage.

Step 1: Add "wavelet_downscaling" folder to your current MATLAB path.

Step 2: Load "sample data.mat". This file contains all the required dataset for the trial run. 

ncep_data= NCEP data,

gcm_data_val= GCM validation data,

gcm_data_sim= Future GCM data,

target= model target data,

imd_chk= model validation data

Detailed data description is available through this schematic for better understanding.

Step 3: Specify model requirements like variable averaging ( Required or not), number of clusters of input variables to be formed based on multi-scale wavelet entropy (MWE) and K-means clustering, allowed percentage variance explained by the principal components for every cluster.

variable_averaging= "1"== YES; "0"== No

variance_limit= range=(0,1] ;eg. 0.9=90%; max value=1 pertaining to 100% explained variance by principal components,

num_clus= number of clusters of input variables to be formed before application of MWE and K- means

Step 4: Run the function.Type the following in MATLAB command window: 

[validation_WP_MLR,validation_P_MLR]=wavelet_downscaling(ncep_data,gcm_data_val,gcm_data_sim,target,imd_chk,1,.9,5);

Here, variable averaging is permitted, 90% of variance is explained

through PCA's and 5 clusters of input variables have been allowed. User can change these values as per the model and data requirement.