Research Article
A Classifier Graph Based Recurring Concept Detection and Prediction Approach
| Algorithm 1: Distribution-Based Detection Method | | Input: S: data stream, m: size of block, , ; | | Output: flagWarning; | | begin | | flagWarning = false; | | for each instance in S do | | numInstance++; | | if numInstance % m == 0 then | | for each check point Pi do | | compute mean values and ; | | compute and using Šidák correction; | | compute and ; | | if then | | delete instances which are on the left side of Pi; | | if then | | return true; | | break; | | end if | | if then | | flagWarning = true; | | end if | | end if | | end for | | return flagWarning; | | end. |
|