Method for determining a preferred node in a classification and regression tree for use in a predictive analysis
AssigneeInternational Business Machines Corporation
InventorsKARLA K. ARNDT, JAMES M. CAFFREY, ASPEN L. PAYTON
Life of this patent
Estimating the term…
Abstract
Techniques are described for determining what node of a classification and regression tree (CART) should be used by a predictive analysis application. A first approach is to use a standard deviation of the data at a given the level of the CART to determine whether data in the next, lower node is more consistent than the data in the current node. A second approach is to measure a correlation between data points in a given node and the time at which each point was sampled (or other correlation metric) to identify a preferred node.
Claims9
A computer-implemented method for determining a set of data to use for a predictive analysis, the method comprising:
- identifying a sequential set of sampled values of at least one performance metric;
- generating, from the set of sampled values, a classification and regression tree, wherein each level of the classification and regression tree below a root node splits the set of sampled values into at least two nodes, each node containing a subset of the sampled values;
- traversing a specified branch of the classification and regression tree until identifying one of the nodes for use by a predictive modeling analysis.
Dependencies are inferred from the claim text.