ModalMap: Modeling and Map production using Random Forest and Stochastic Gradient Boosting

### before you begin ###

If you are installing from a local zip file

Make certain you have already installed packages: gbm, randomForest, rgdal, raster, PresenceAbsence, fields, HandTill2001

Install ModalMap package using Packages > Install package(s) from local zip files...

### Example datasets ###

After installing ModelMap, there will be a subdirectory extdata under the package dirrectory in your installation.

For Example: C:\R\R-2.6.2\library\ModelMap\extdata 

This contains the example datasets for the help examples and for the vignette examples.

### Help Example dataset ###

The help example dataset contains Training and Test data .csv files with 8 columns

An index variable to identify each row:

     * ID

Three response variables:

     * BIO        Continuous response variable of above ground Biomass
     * BIOCAT     Categorical response variable
     * CONIFTYP   Binary response variable of presence/absence of Conifer Forest Types

Four predictor variables:

     * 3 tasselcap (TC) bands as continuous predictor variables 
     * NLCD (National Land Cover Dataset) as a catagorical predictor variable

The .csv files LUT_2001.csv and LUT_2004.csv are look up tables for associating the column headers in the training and test datasets
with the image files for map production.

There is one image file for NLCD data.
There are two multiband image files for tasselcap data giving 2001 data and 2004 data.

### Vignette Example dataset ###

The help example dataset contains a training data file: "VModelMapData.csv" with 38 columns (though not all columns are used in the vignette)

The vignette uses:

An index variable to identify each row:

     * ID

Response variables:

     * PINYON and SAGE   Continuous response variables of percent cover of Pinyon and Sage
     * VEGCAT            Categorical response variable

For Binary Presence/Absence models, the percent cover of PINYON and SAGE are transformed to a 0/1 variable, where any value greater than 0 is counted as presence.

Six predictor variables:

Continuous predictor variables:

     * ELEV250
     * EVI2005097
     * NDV2005097
     * NIR2005097
     * RED2005097

Factored predictor variable:

     * NLCD01_250


The .csv file "VModelMapData_LUT.csv" is a look up table for associating the column headers in the training dataset
with the image files for map production.

There is one image file for National Landcover Class Dataset.
There is one image file for elevation.
There is one multiband image file for remote sensing data.