Braincube designed new data mining technology for scoring the impact of input data to output data, using a probabilistic approach : the CrossRank.
The CrossRank scoring is reliable in any situation and the answer is the best range of impact.
The Braincube CrossRank is a data mining scoring tool used to quantify the impact of an Input variable on an Output variable, with a scoring between 0 to 100.
The output is split into « good » points (the analyst targets a range) versus « bad » points. The CrossRank evaluates how much an input variable can separate the goods points from the bad points of the output.
A value of “100” means that the variable can separate all good points from all bad ones.
A value of “0” means that the variable can't be more effective than a random variable.
Computing a dataset with the CrossRank corresponds to all possible intervals (I) of the input variables : the CrossRank final value is the value of the best range. The CrossRank methodology gives the analyst reliable scoring, and it also identifies the most impactful range of the input variable on the output variable.
To go deeper, you can read the document in the Downloads tab on the left.