Algorithm SMOTE (S, M, A)

Input:

A= Number of nearest neighbors

M= Number of samples from minority classes

S= Amount of SMOTE S %

Output: ($\frac{S}{100}$) * M Samples of synthetic minority classes

(“Identify minority class samples in the data set. If S is less than 100%, randomize the

minority class samples since only a random percentage of them will be SMOTE”)

ifS <100

Then determine the minority class sample M randomly

M = ($\frac{S}{100}$) ∗M

S= 100

endif

S= (int)($\frac{S}{100}$) (“determine the number of SMOTE samples assuming that the number

is an integer with a multiple of 100”)

A= Number of nearest neighbors

numattrs= Number of attributes

Sample [ ] [ ]: is a number of original samples in the form of Arrays from the minority

class

newindex: the resulting number of samples is calculated by initializing to 0

Synthetic [ ] [ ]: Array for synthetic samples (“only calculates the number of nearest

neighbors of A for each sample”)

for i← 1 toM

nearest neighbor i is calculated using A then store index into array n

Populate (S, i, n array)

endfor

Populate (S, i, nnarray) (“this is a manufacturing function with synthetic samples”)

whileS≠ 0

select data one from “A” for nearest neighbor “i” by randomly selecting 1 and A

for attr← 1 tonumattrs

Compute: dif=Sample [nnarray [nn ]] [attr ] −Sample [i ][attr ]