On Software Defect Prediction Using Machine Learning

Algorithm 2

AKPCAC.

Input: The set of Labeled samples, 𝐿=𝑥₁,𝑦₁,𝑥₂,𝑦₂,…,𝑥_ℓ,𝑦_ℓ;

The set of unlabeled samples, 𝑈={𝑥_ℓ+1,𝑥_ℓ+2,…,𝑥_ℓ+𝑢};

Output: Kernel Principal Component Classifier, 𝐻;

Method:

(1) 𝐾_𝑖𝑗=𝜅(𝑥_𝑖,𝑥_𝑗),𝑖,𝑗=1,…,ℓ;

(2) 𝐾=𝐾−(1/ℓ)𝐽𝐽^′𝐾−(1/ℓ)𝐾𝐽𝐽^′+1/ℓ²(𝐽^′𝐾𝐽)𝐽𝐽^′, where 𝐽 is a vector with all elements equal to 1.

(3) [𝑉,Λ]=eig(𝐾);

(4) 𝛼=∑^𝑝_𝑗=1(1/𝜆_𝑗)(𝑉^′_𝑗𝑌^𝑠)𝑉_𝑗; %𝑌^𝑠=𝑦₁,𝑦₂,…,𝑦_ℓ is the label vector

(5) Calculate 𝑏̂,𝐻(𝑥) according to (9), (10);

(6) return   𝐻;

End Algorithm APPCC.