Customer Relationship Management Based on SPRINT Classification Algorithm under Data Mining Technology

Algorithm 1

SPRINT algorithm

(i)	{
(ii)	Preprocess data set S, establish list C based on attributes in A, create root node R, and create junction
(iii)	Point queue Q.
(iv)	While (queue Q is not empty)
(v)	{
(vi)	Fetch node N from queue Q
(vii)	If (records in node N's attribute list are of the same category or few)
(viii)	then
(ix)	{
(x)	Mark node N as a leaf node.
(xi)	The continue
(xii)	}
(xiii)	33
(xiv)	Each split point K in for N
(xv)	{
(xvi)	Create a class histogram
(xvii)	The Gini index of the split point K is calculated to select the best K '.
(xviii)	}
	According to K ′, node N is divided into two children, and N1 and N2 are put into the queue.
	Divide the list where K ′ is located, create a hash table according to the RIds of the list, and divide other attribute lists.
	}
	MDL algorithm is used to prune decision tree.
	}
	According to the theory of marketing, the purchasing power of customers is very important [29], which has a great influence on the formulation of marketing strategies by enterprises, so the target variable is customer purchasing power. Customer purchasing power is divided into three types: high purchasing power, average purchasing power, and low purchasing power. Customers are defined according to their age, gender, occupation, family status, and other attributes. A customer information table is arranged in order according to customer information, as shown in Table 1.