RAI Random Forest
Click to view larger images
✓ Compatible with desktop and web
Random Forest is a very popular and very powerful machine learning algorithm. The run times are quite fast and it is able to deal with data sets that are multi-dimensional, unbalanced, and contain missing values.
- Load in your own multidimensional data set
- Remove outliers or filter out unwanted data points
- Generate Random Forest models with a click of a button
- Analyze the results that include predicted values, model fit stats, and Variable Importance measures
- Simple, easy-to-read TERR code which can be modified by the user
RAI Random Forest template supports multidimensional data sets, with both categorical or continous dimensions. The data set can include missing values as the template incorporates the imputation functionality.
In the analysis workflow, the data points that have populated response variables are included in the training set and are used to calculate the goodness of fit measures (Out of Bag Error, % variance explained for Regression Forest). Those records, where the response variable is empty, will have the values predicted and will not be included in calculations of the model fit statistics.
- RandomForest – grows the random forest based on selected dimensons and response variable; uses the CRAN R library randomForestSRC 2.1.0. https://cran.r-project.org/web/packages/randomForestSRC/index.html
- InstallPackages – installs the necessary R libraries
- MarkOutliers – marks the marked records as outliers
- GetColumnName – extracts the column names of the Input Table and cretes an Output Table that consists of one column with those names; those names are considered the possible dimensions of the data.
The main Random Forest method uses the CRAN R library randomForestSRC 2.1.0. It grows the random forest (Breiman, 2001) using user supplied training data. It can be applied to the selected response (outcome) that is either numeric or categorical (factor) and yields regression and classification forests, respectively.
The package implements OpenMP shared-memory parallel programming. The instructions on how to set the JAVA_HOME Variable necessary to use the multicore functionality are included in the template.
A template is an analysis file created in TIBCO Spotfire. The intended use is to replace the data and utilize the configured visualizations and calculations.
- Download the template to your personal machine. Login into your own Spotfire server and open the downloaded file.
- Review the tables shown on the data load page. These are the tables you'll replace.
- Replace the data with another Excel sheet or connect the template directly to your database or platform.
- Customize with your own data. You are free to embed and incorporate your template into other projects.
- All done!
Spotfire Templates are provided as is and do not require licenses.
RAI Random Forest
- Your cart is empty.
|Release date:||June 1, 2016|
|Last updated:||June 1, 2016|
|Software application type:||Spotfire Template|
|File format:||.dxp, .zip|
|Requirements:||TIBCO Spotfire 7.0+ Professional|