Algorithm 1 Pseudocode of Dataset Sampling

Input: A list of $N-\mathit{BaIoT} \mathrm{files} F$

Output: Balanced dataset

$DF\leftarrow$ an empty list

$s\leftarrow$ size of data frame

for each file $f\in F$ do:

Import the file $f$ as data frame $df$

Count the size $s^{df}$ of the $df$

Append data frame to $df\in DF$

End for

threshold $\theta \leftarrow$ the smallest data frame size $s\left(df\right)$

For all$df\in DF$ AND $s\left(df\right)>\theta$ do:

$\hspace{1em}p\leftarrow$ percent of data $\%$

Sample dataset as $d{f}_i\leftarrow \left(\frac{\theta }{df}\right)*100\%$

End For

Return $d{f}_i$ as csv format

End