The Statistical and Predictive Analytics node pack provides a number of nodes that enable you to use statistical and analytic techniques without needing to use R language coding. These nodes are separately licensed and leverage the embedded TIBCO® Enterprise Runtime for R software.
Prerequisites
Ensure that you have the required Statistical and Predictive Analytics node pack license. You can apply a new license at any time, see Applying a new license.
Nodes in the Statistical and Predictive Analytics node pack depend on functionality provided by open source licensed CRAN packages. On Windows you can download and install the required CRAN R library packages by using the R Library Package Download node.
Due to the nature of the official CRAN updates and potential incompatibility issues, Precisely hosts a repository with tried and tested versions of the packages. We only host those packages known to be compatible with the Statistical and Predictive Analytics node pack (powered by TIBCO®). If you want to install other packages, this must be done on the understanding that the latest versions may not be compatible.
On Windows you can download and install the required CRAN R library packages by using the R Library Package Download node.
To install the packages on Linux the following steps are required:
- Install the open-source R product as described at: https://www.r-project.org/. The supported version of R required by the Precisely repository is 4.3.0. Precisely cannot guarantee compatibility with later versions.
- Open the R console and run the following routine:
install.packages(c("arules","boot","cluster","lattice","Matrix","MASS","NbClust","quantreg","randomForest","SparseM","survival","xts","zoo"),destdir="<site-dir>/tmp/TERR/downloads",contriburl="http://cran.lavastorm.com/src/contrib_4.3",type="source",lib="<site-dir>/tmp/TERR/libs", method="libcurl")Note that you should replace the <site-dir> placeholder in the “destDir” and “lib” properties to point to your site configuration location.
Please ensure that the folders referenced in the destdir and lib properties in the above command, exist before running the command.
Nodes
The following table gives an overview of the nodes that are available in the Statistical and Predictive Analytics node pack:
| Nodes | Description | Example |
|---|---|---|
|
|
Used to predict outcomes by repeatedly identifying patterns from an existing data set. | Is a particular customer with a certain set of attributes likely to switch to a competitor? |
| Hierarchical Clustering | Used to form a user-specified number of clusters out of data sets using user-defined criteria based on an iterative process of cluster merging. | Segment insurance policy holders into groups based on expected claims costs. |
|
|
Used to form a user-specified number of clusters out of data sets using user-defined criteria based on proximity. | Segment markets for differentiated pricing. |
|
|
Used to calculate a line or curve of best fit to estimate values. | How much revenue can be generated for each dollar spent on advertising? |
|
|
Used to calculate probabilities of binary outcomes. | What is the probability that a customer will make a purchase, given certain customer attributes? |
|
|
Used to discover co-occurrence relationships between transaction items, activities, or characteristics. | Based on the TV shows that a viewer has watched, what other shows might the viewer also enjoy? |
|
|
Used to calculate an outcome of interest given a specific quantile (or percentile) and a specific attribute value. | How responsive are students to different education factors, given a certain standardized test score? |
|
|
A method that predicts future values of interest using weighted past values. | What is the expected value of retail sales for December? |