Incremental Optimization Mechanism for Constructing a Decision Tree in Data Stream Mining

Pseudocode 1

Pseudocode of input and the test-then-train approach.

INPUT:

𝑆: A stream of sample

𝑋: A set of symbolic attributes

𝐻(⋅): Heuristic function using for node-splitting estimation

𝛿: One minus the desired probability of choosing a correct attribute at any given node

𝑛_min: The minimum number of samples between check node-splitting estimation

ℱ: A functional tree leaf strategy

OUTPUT:

HT: A decision tree

PROCEDURE: OVFDT (𝑆, 𝑋, 𝐻(⋅), 𝛿, 𝑛_min, ℱ)

(1) A data stream 𝑆 arrives

(2) IF HT is null, THEN  initializeHT(𝑆, 𝑋, 𝐻(⋅), 𝛿, 𝑛_min, ℱ)

 ELSE  traverseHT(𝑆, 𝐻𝑇, ℱ) and update Δ𝐶

(3) Label 𝑙 as the predicted class among the samples seen so far

(4) Let 𝑛_𝑙 be the number of samples seen at the leaf 𝑙

(5) IF the samples seen so far at leaf 𝑙 do not all belong to the sameclass

  and (𝑛_𝑙 mod 𝑛_min) is zero, THEN doNodeSplitting(Δ𝐶, 𝑆, 𝑋, 𝐻(⋅), 𝛿, 𝑛_min)

(6) Return HT