Concurrent data predictor

Abstract

A set of methods and corresponding systems that analyze, predict, or classify data. The improvements are the result of techniques that leverage the concurrent evaluation of attributes in the reference data set entries.

What is claimed is: 
     
         1 . A computer-implemented method for estimating the likelihood of unknown outcomes comprising:
 accessing a reference data structure containing rows, wherein each row includes a target outcome and column values corresponding to feature attributes;   accessing a query data entry containing column values corresponding to attributes of said reference data structure; and   calculating a likelihood measure of said target outcomes, said calculation comprising:
 evaluating rows of the reference data structure and processing each evaluated row, where the processing comprises:
 evaluating each of said attributes, and for each pair of values, one from the query data entry and one from the current row, computing a match column score; 
 computing an entry match score for the ensemble of said match column scores; and 
 updating the work data set entry corresponding to said entry match score with the current target outcome; and 
 
 extracting likelihood measures for target outcomes by processing said work data set. 
   
     
     
         2 . The method of  claim 1 , wherein said match column score is a function of match degree, features of the attribute, and features of the column values in said pair. 
     
     
         3 . The method of  claim 2 , wherein said match column score favors columns with rare column values. 
     
     
         4 . The method of  claim 1 , wherein said reference data structure is dynamically updated. 
     
     
         5 . The method of  claim 4 , wherein said dynamic update consists in associating a timestamp to the rows of said reference data, structure and removing the oldest rows when a capacity threshold for storing rows is reached. 
     
     
         6 . A computer-implemented method for estimating the likelihood of unknown outcomes comprising:
 accessing a reference data structure containing rows, wherein each row includes a target outcome and column values corresponding to feature attributes;   accessing a query data entry containing column values corresponding to attributes of said reference data structure; and   calculating a likelihood measure of said target outcomes, said calculation comprising:
 evaluating rows of the reference data structure and processing each evaluated row, where the processing comprises:
 evaluating each of said attributes, and for each pair of values, one from the query data entry and one from the current row, computing a match column score; 
 computing an entry match score for the ensemble of said match column scores; and 
 updating the work data set entry corresponding to the current target outcome with said entry match score; and 
 
 extracting likelihood measures for target outcomes by processing said work data set. 
   
     
     
         7 . The method of  claim 6 , wherein said match column score is a function of match degree, features of the attribute, and features of the column values in said pair. 
     
     
         8 . The method of  claim 7 , wherein said match column score favors columns with rare column values. 
     
     
         9 . The method of  claim 6 , wherein said reference data structure is dynamically updated. 
     
     
         10 . The method of  claim 9 , wherein said dynamic update consists in associating a timestamp to the rows of said reference data structure and removing the oldest rows when a capacity threshold for storing rows is reached.