Customer behaviour analysis based on buying-data sparsity for multi-category products in pork industry: A hybrid approach

Figures & data

Table 1. Example of five customers based on RFM values for overall and each product categories

Customer no.	Ham	Tenderloin	Loin	Collar	Shoulder	Belly	Spare rib	By-product	Trimming	Overall values
Product Category (Recency values, Avg.)
N01	3	-	-	-	9	24	33	-	10	3
N02	39	-	-	-	-	-	47	2	2	2
N03	-	-	-	-		-	-	9	-	9
N04	1	-	-	-	1	1	-	1	1	1
N05	-	-	16	-	-	-	16	2	-	2
Product Category (Frequency values, Avg.)
N01	9	-	-	-	4	2	2	-	3	15
N02	1	-	-	-	-	-	1	18	2	20
N03	1	-	-	-	-	-	-	4	-	4
N04	18	-	-	-	6	6	-	6	8	19
N05	-	-	1	-	-	-	1	2	-	2
Product Category (Monetary values, Avg.)
N01	6,139	-	-	-	17,203	1,488	1,250	-	3,600	179,200
N02	1,345	-	-	-	-	-	2,500	686	661	20,228
N03	-	-	-	-	-	-	-	800	-	5,600
N04	2,989	-	-	-	1,026	1,576	-	460	1037	163,962
N05	-	-	7,092	-	-	-	1,500	400	-	9,392

Figure 1. Proposed methodology

Figure 2. Graph of the aspect model

Figure 3. AIC values versus the number of clusters

Table 2. RFM value criterion for all product categories

Product				Scaling
Category	Value	1 score	2 score	3 score	4 score	5 score
	R	40 up	26–40	17–25	5–16	5 down
Ham	F	5 down	5–15	16–30	31–40	40 up
	M	700 down	700–1,200	1,201–3,000	3,001–10,000	10,000 up
	R	30 up	21–30	11–20	5–10	5 down
Tender loin	F	10 down	11–15	16–25	26–40	40 up
	M	400 down	400–800	801–1,500	1,501–3,000	3,000 up
	R	30 up	21–30	11–20	5–10	5 down
Loin	F	5 down	5–15	16–25	26–40	40 up
	M	800 down	800–1,800	1,801–3,500	3,501–6,000	6,000 up
	R	30 up	21–30	11–20	5–10	5 down
Collar	F	5 down	5–15	16–25	26–40	40 up
	M	800 down	800–1,600	1,601–3,000	3,001–5,000	5,000 up
	R	30 up	21–30	11–20	5–10	5 down
Shoulder	F	5 down	5–15	16–30	31–40	40 up
	M	600 down	600–1,200	1,201–3,500	3,501–20,000	20,000 up
	R	30 up	21–30	11–20	5–10	5 down
Belly	F	5 down	5–15	16–25	26–40	40 up
	M	700 down	700–1,500	1,501–4,000	4,001–10,000	10,000 up
	R	30 up	19–30	11–18	5–10	5 down
Spare rib	F	5 down	5–15	16–25	26–40	40 up
	M	500 down	500–1,300	1,301–2,500	2,501–5,000	5,000 up
	R	30 up	21–30	11–20	5–10	5 down
By-product	F	5 down	5–15	16–25	26–40	40 up
	M	200 down	200–500	501–1,000	1,001–5,000	5,000 up
	R	30 up	21–30	11–20	5–10	5 down
Trimming	F	5 down	5–15	16–25	26–40	40 up
	M	600 down	600–1,000	1,001–3,500	3,501–10,000	10,000 up

* R (days); F (number of times); M (Thai Baht)

Table 3. P(Z_k|C_i) matrix elements for 9 customer clusters

	Cluster number
Customer no.	1	2	3	4	5	6	7	8	9	Sum
N01	0.00	0.10	0.24	0.13	0.00	0.00	0.00	0.29	0.24	1.00
N02	0.00	0.60	0.00	0.00	0.00	0.00	0.00	0.00	0.40	1.00
N03	0.00	0.00	0.00	0.00	0.00	1.00	0.00	0.00	0.00	1.00
N04	0.00	0.00	0.20	0.20	0.00	0.18	0.00	0.22	0.18	1.00
N487	0.00	0.45	0.55	0.00	0.00	0.00	0.00	0.00	0.00	1.00

Table 4. Customer segments based on RFM values of the all product category

Ham product category
Cluster no.	No. of cust.	Recency (Avg.)			Frequency (Avg.)			Monetary (Avg.)
		Unit time	R		No.	F		Unit cost	M
1	58	6.88	4.48	↑	21.95	2.88	↑	4,965.03	2.50	↓
2	12	10.25	4.08	↓	11.25	2.00	↓	1,675.37	2.33	↓
3	12	15.58	3.58	↓	7.42	1.67	↓	3,117.34	2.17	↓
4	13	11.08	4.00	↓	9.62	1.92	↓	20,039.58	3.62	↑
5	0	-	-	-	-	-	-	-	-	-
6	14	11.79	4.07	↓	9.21	1.86	↓	4,713.42	2.50	↓
7	7	10.57	4.00	↓	5.29	1.29	↓	1,107.17	2.14	↓
8	55	8.07	4.36	↑	14.65	2.35	↑	49,755.36	3.87	↑
9	10	3.20	4.80	↑	13.80	2.10	↑	31,656.84	3.10	↑
	Avg.	-	4.17	-	-	2.01	-	-	2.78	-
Tenderloin product category
Cluster no.	No. of cust.	Recency (Avg.)			Frequency (Avg.)			Monetary (Avg.)
		Unit time	R		No.	F		Unit cost	M
1	92	4.49	4.62	↑	21.66	2.83	↑	2,334.32	3.15	↑
2	10	23.20	2.80	↓	3.00	1.10	↓	844.96	2.10	↓
3	7	8.57	4.14	↑	7.86	1.43	↓	505.57	1.57	↓
4	7	17.71	3.29	↓	5.00	1.14	↓	1,125.19	2.14	↓
5	5	29.40	1.80	↓	1.80	1.00	↓	1,010.11	2.80	↑
6	4	13.75	3.50	↓	8.50	1.50	↓	275.51	1.25	↓
7	4	6.25	4.25	↑	6.25	1.25	↓	882.13	2.25	↑
8	4	15.25	3.75	↑	12.50	1.75	↑	1,805.47	3.00	↑
9	5	2.40	4.80	↑	12.80	2.00	↑	625.39	1.80	↓
	Avg.		3.66			1.56			2.23
Loin product category
Cluster no.	No. of cust.	Recency (Avg.)			Frequency (Avg.)			Monetary (Avg.)
		Unit time	R		No.	F		Unit cost	M
1	90	4.48	4.59	↑	22.49	2.98	↑	3,419.13	2.60	↑
2	20	17.60	3.15	↓	10.40	1.95	↑	1,529.67	2.00	↓
3	18	10.33	3.94	↓	8.11	1.72	↓	1,296.70	1.94	↓
4	15	11.80	3.60	↓	4.07	1.40	↓	1,953.31	2.40	↓
5	98	9.37	3.96	↓	7.17	1.69	↓	6,445.54	3.58	↑
6	9	6.78	4.44	↑	8.11	1.88	↓	1,124.36	1.78	↓
7	23	11.91	3.65	↓	5.39	1.48	↓	3,356.05	3.00	↑
8	17	11.71	3.71	↓	6.41	1.53	↓	4,843.37	2.88	↑
9	10	3.60	4.70	↑	17.70	2.40	↑	16,485.80	2.90	↑
	Avg.		3.97			1.89			2.57
Collar product category
Cluster no.	No. of cust.	Recency (Avg.)			Frequency (Avg.)			Monetary (Avg.)
		Unit time	R		No.	F		Unit cost	M
1	61	2.66	4.85	↑	27.70	3.41	↑	3,982.51	2.61	↓
2	11	16.00	3.55	↓	10.55	2.00	↓	2,337.78	2.27	↓
3	15	10.87	3.87	↓	11.20	2.00	↓	1,733.41	2.27	↓
4	14	8.79	4.07	↓	7.93	1.71	↓	3,084.61	2.64	↑
5	9	8.11	4.22	↑	5.33	1.44	↓	1,935.49	2.56	↓
6	5	2.40	5.00	↑	8.20	1.60	↓	1,394.66	2.00	↓
7	48	9.04	4.02	↓	9.08	1.88	↓	6,223.52	3.83	↑
8	8	11.88	3.63	↓	9.00	1.75	↓	6,216.50	3.00	↑
9	7	9.43	4.00	↓	13.57	2.29	↑	4,647.78	2.57	↓
	Avg.		4.13			2.01			2.64
Shoulder product category
Cluster no.	No. of cust.	Recency (Avg.)			Frequency (Avg.)			Monetary (Avg.)
		Unit time	R		No.	F		Unit cost	M
1	32	7.53	4.13	↓	17.47	2.47	↑	5,751.99	2.38	↓
2	5	14.80	3.60	↓	5.40	1.40	↓	4,322.15	3.00	↑
3	7	3.86	4.71	↑	13.43	2.00	↑	58,655.06	2.43	↑
4	8	5.13	4.50	↑	16.25	2.50	↑	784.89	1.88	↓
5	1	2.00	5.00	↑	4.00	1.00	↓	524.98	1.00	↓
6	9	13.56	3.44	↓	4.22	1.33	↓	1,826.06	2.22	↓
7	2	2.50	5.00	↑	6.50	1.50	↓	643.90	1.50	↓
8	14	8.36	4.21	↓	13.29	2.07	↑	26,689.40	3.50	↑
9	22	7.05	4.45	↑	15.82	2.36	↑	22,591.54	3.82	↑
	Avg.		4.34			1.85			2.41
Belly product category
Cluster no.	No. of cust.	Recency (Avg.)			Frequency (Avg.)			Monetary (Avg.)
		Unit time	R		No.	F		Unit cost	M
1	69	4.14	4.67	↑	22.97	3.04	↑	3,325.44	2.06	↓
2	11	18.45	2.91	↓	6.64	1.64	↓	3,606.94	2.27	↓
3	19	15.11	3.32	↓	7.37	1.58	↓	1,410.48	1.84	↓
4	51	6.63	4.33	↑	12.75	2.12	↑	18,455.27	3.37	↑
5	15	10.00	3.87	↓	5.73	1.47	↓	2,168.35	1.93	↓
6	12	3.33	4.83	↑	6.50	1.50	↓	1,406.34	2.08	↓
7	14	4.07	4.64	↑	9.79	2.00	↑	3,409.42	2.86	↑
8	17	10.00	3.88	↓	10.47	1.88	↓	5,564.33	2.71	↑
9	9	1.33	5.00	↑	17.44	2.67	↑	23,300.72	3.11	↑
	Avg.		4.16			1.99			2.47
Spare rib product category
Cluster no.	No. of cust.	Recency (Avg.)			Frequency (Avg.)			Monetary (Avg.)
		Unit time	R		No.	F		Unit cost	M
1	67	3.78	4.72	↑	25.46	3.24	↑	2,771.56	2.42	↑
2	35	13.43	3.63	↓	10.34	1.86	↓	4,121.64	3.00	↑
3	16	16.06	3.44	↓	10.06	1.81	↓	1,375.82	2.13	↓
4	8	9.88	4.00	↑	14.75	2.38	↑	2,745.68	2.38	↑
5	5	8.40	4.00	↑	3.40	1.40	↓	1,415.16	2.40	↑
6	18	15.17	3.50	↓	10.17	2.06	↓	2,892.43	1.94	↓
7	10	13.60	3.60	↓	10.70	2.10	↑	1,123.13	2.20	↓
8	18	12.22	3.56	↓	8.61	1.78	↓	1,415.38	2.39	↑
9	9	7.44	4.11	↑	10.78	2.00	↓	1,724.79	2.44	↑
	Avg.		3.84			2.07			2.37
By-product product category
Cluster no.	No. of cust.	Recency (Avg.)			Frequency (Avg.)			Monetary (Avg.)
		Unit time	R		No.	F		Unit cost	M
1	50	4.34	4.70	↑	26.34	3.34	↑	570.73	1.70	↓
2	6	5.83	4.33	↑	11.17	2.17	↑	334.65	1.83	↓
3	18	5.44	4.56	↑	10.28	1.89	↓	479.64	2.22	↑
4	12	14.83	3.42	↓	9.83	1.92	↓	225.56	1.42	↓
5	5	7.00	4.40	↑	4.40	1.40	↓	445.40	2.00	↓
6	37	8.27	4.16	↓	13.73	2.27	↑	10,889.97	3.22	↑
7	9	12.56	3.56	↓	9.56	1.89	↓	1,661.73	2.44	↑
8	19	12.47	3.89	↓	11.05	2.16	↑	4,473.26	2.58	↑
9	8	3.00	4.75	↑	14.38	2.38	↑	402.58	2.13	↓
	Avg.		4.20			2.16			2.17
Trimming product category
Cluster no.	No. of cust.	Recency (Avg.)			Frequency (Avg.)			Monetary (Avg.)
		Unit time	R		No.	F		Unit cost	M
1	63	6.06	4.41	↑	21.52	2.87	↑	3901.30	2.24	↓
2	19	14.95	3.47	↓	8.68	1.74	↓	1715.63	2.37	↓
3	49	5.57	4.45	↑	14.37	2.33	↑	5475.34	3.45	↑
4	20	9.90	4.00	↓	11.25	2.05	↑	2333.82	2.15	↓
5	12	10.58	3.75	↓	9.83	1.75	↓	1278.19	2.08	↓
6	22	13.45	3.59	↓	9.41	1.73	↓	1589.63	2.59	↑
7	13	9.92	3.92	↓	8.92	1.69	↓	2583.91	2.85	↑
8	26	10.85	3.81	↓	8.85	1.73	↓	5449.34	2.65	↑
9	9	4.11	4.67	↑	17.33	2.56	↑	3951.37	2.22	↓
	Avg.		4.01			2.05			2.51

Table 5. Customer segments for all of the product categories

Cluster no.	Product Category
	Ham	Tenderloin	Loin	Collar	Shoulder	Belly	Spare rib	By-product	Trimming
1	SH	VC	VC	SH	FC	SH	VC	SH	SC
2	UC	UC	FC	UC	SP	UC	SP	SH	UC
3	UC	NC	UC	UC	VC	UC	UC	PC	VC
4	SP	UC	UC	SP	SH	VC	VC	UC	FC
5	-	SP	SP	NC	NC	UC	PC	NC	UC
6	UC	UC	NC	NC	UC	NC	UC	CC	SP
7	UC	PC	SP	SP	NC	VC	UC	SP	SP
8	VC	VC	SP	SP	CC	SP	SP	CC	SP
9	VC	SP	VC	FC	VC	VC	PC	SP	SP

Table 6. Customer segments without considering product categories

Cluster no.	No. of cust.	Recency (Avg.)		Frequency (Avg.)		Monetary (Avg.)		RFM pattern			Cust. type
		Unit time	R	No.	F	Unit cost	M	R	F	M
1	92	4.17	4.63	23.92	3.03	20,944.74	4.23	↑	↑	↑	VC
2	35	13.46	3.63	11.63	2.00	9,286.07	3.74	↓	↓	↓	UC
3	49	5.39	4.47	14.80	2.31	16,869.63	3.86	↑	↑	↓	SH
4	51	6.53	4.35	13.65	2.20	26,661.21	4.02	↑	↓	↓	NC
5	98	9.20	3.99	7.68	1.70	7,263.52	3.77	↓	↓	↓	UC
6	37	8.05	4.19	15.38	2.38	16,417.77	3.62	↓	↑	↓	FC
7	48	9.00	4.04	10.00	1.90	10,333.21	4.25	↓	↓	↑	SP
8	55	7.42	4.31	15.49	2.38	65,386.06	4.36	↑	↑	↑	VC
9	22	6.41	4.45	18.86	2.64	58,096.09	4.45	↑	↑	↑	VC
Total	487		4.23		2.28		4.03

Table 7. Examples of customer orders

Customer	Cluster no.	Ham	Loin	Shoulder	Total
N01	1	60	150	150	360
N02	3	-	200	-	200
N03	5	-	-	300	300
N04	7	100	100	200	400
N05	8	240	-	-	240
N06	9	200	100	300	600
N07	9	-	350	250	600
Total (pieces)	-	600	900	1,200	2,700