Prediction of Repeat Customers on E-Commerce Platform Based on Blockchain

Algorithm 1

The sub-time under sampling sample balancing algorithm.

Input: The original historical data about buyers (); the numbers of recording days ();

Output: The balanced historical data about buyers ().

1: ; //Segment the Original Data according to the Number of Recording Days

2: fordo//Traverse each buyer of the original data.

3: = Random Choose()//Randomly select a buyer sample.

4: if is a repeat buyer then.

5: //Determine whether there are more than two one-time buyers in the nearest neighborhood of .

   if repeat-buy = sum(KNN()) then.

6: //Remove the One-Time Buyers from the Nearest Neighbors

delete (KNN() repeat-buy).

7:  else.

8:    save ()//Keep this buyer sample.

9: else.

10:  if norepeat-buy = sum(KNN()) then.

11:   delete ()//Remove this buyer sample.

12:  else.

13:   save ()//Keep this buyer sample.