SAND’s column database delivers wide ranging support for the Data Mining Group‘s open PMML standard (Predictive Model Markup Language), providing advanced analytics on big data. Complex mathematical and statistical models, the creation of advanced User Defined Functions, the elimination of extraction requirements, and the ability to execute analytics directly on the data are just some of the benefits offered via SAND’s industry-leading PMML support.
- Association Rules, including recommendation, exclusiveRecommendation, ruleAssociation
- Clustering Model, supporting center-based and distribution-based models, and all kinds of distances: euclidean, squaredEuclidean, chebychev, cityBlock, minkowski, simpleMatching, jaccard ,tanimoto, binarySimilarity.
- Regression Model, including linearRegression, stepwisePolynomialRegression, logisticRegression
- Neural Network Model, which supports activation functions such as: threshold, logistic, tanh, identity, exponential, reciprocal, square, Gauss, sine, cosine, Elliott, arctan
- Naïve Bayes
- Support Vector Machine, including Kernel Types: LinearKernelType, PolynomialKernelType, RadialBasisKernelType, SigmoidKernelType
- Ruleset Model
- Tree Model
- Functions: Normalization, Discretization, Mapping Values, Aggregation, User Defined Functions, Some elements of MathML, Matrix Operations, and the following built-in functions.
SAND’s PMML implementation delivers easy, standards-based models imported from SAS, SPSS, KXEN, or any PMML producer, applying predictive modeling and in-database analytics on billions of rows without moving the data in or out, and executing to meet user requirements.
Learn more
To receive more information about SAND PMML support, please complete and submit the following form:
