Table of Contents
Security polices in corporations usually disallow inclusion of third party software with unknown source code into the core system. Such software cannot be considered fully secure or efficient enough. It is more practical and safe to generate source code which can be used without the Data Mining software. It is much more easier to examine such small piece of code than the whole Data Mining system. This type of source code is called scoring code.
AdvancedMiner provides scoring code generation for Data Mining models and data transformations. AdvancedMiner generates scoring code only in the Java programming language.
The generated code is stand-alone, i.e. it does not need AdvancedMiner and its libraries to work. In fact, the compilation and execution of scoring code results in the same data output as in the case of the execution of Apply Task in AdvancedMiner. Scoring code includes only instructions for data processing and generating model output. The data must be supplied by the programmer.
Scoring code enables the programmer to manually modify:
input attributes mapping
categorical attributes coding (categorical attributes are coded to numerical values e.g. in decision trees)
output data
data preprocessing (transformations)