How does alliance network embedding affect firm innovation? Evidence from the Chinese manufacturing industry

Figure 1.

Figure 2.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Crucial decision rules by replacing random seed (random_state=32)_

Network type	Conditional factor					Criteria		Decision factor
	CB	CI	CC	CL	BC	Support	Confidence	PE
Dyadic network	-	<=25.5	-	-	-	62.86%	0.66	High
	-	>25.5	-	-	-	4.29%	1.00	Low
Star network	-	-	<=0.002	<=0.367		45.39%	0.70	High
	-	-	-	>0.367	-	7.23%	0.80	Low
Ringlike network	-	<=2.583	-	<=0.367	-	4.21%	0.65	Low
	-	(2.583,14]	-	<=0.367	-	8.74%	0.77	High
	-	>14	-	<=0.367	-	3.88%	1.00	Low
	-	-	-	>0.367	-	69.57%	0.89	Low
Complex network	-	-	-	<=0.339	-	38.54%	0.69	High
	-	>2.44	-	>0.339	-	8.52%	0.68	Low
	>2.50	<=2.44	-	>0.339	-	20.53%	0.97	Low

Descriptive statistics and correlation analysis results_

	Mean	Sth	Median	MIN	MAX	PE	CB	CI	BC	CC	CL
PE	2.462	0.758	2.436	0.917	4.472	1.000
CB	2.715	3.190	2.000	1.000	20.000	-0.272***	1.000
CI	6.784	15.000	2.000	0.285	104.28	-0.067***	0.115***	1.000
BC	0.0004	0.002	0.000	0.000	0.011	0.053*	0.586***	0.045*	1.000
CC	0.035	0.036	0.012	0.000	0.100	-0.067**	0.270***	0.002	0.340***	1.000
CL	0.328	0.438	0.000	0.000	1.000	-0.547***	0.205***	-0.077***	-0.101***	0.172***	1.000

Crucial decision rules by using C4_5 model_

Network type	Conditional factor					Criteria		Decision facor
	CB	CI	CC	CL	BC	Support	Confidence	PE
Dyadic network	-	<=23.5	-	-	-	60.16%	0.70	High
	-	>23.5	-	-	-	5.29%	0.98	Low
Star network	-		<=0.002	<=0.371		40.05%	0.77	High
	-	-	-	>0.371	-	8.33%	0.81	Low
Ringlike network	-	<=2.580	-	<=0.371	-	4.55%	0.69	Low
	-	(2.580,13.65]	-	<=0.371	-	6.04%	0.75	High
	-	>13.65	-	<=0.371	-	3.88%	0.98	Low
	-	-	-	>0.371	-	68.85%	0.85	Low
Complex network	-	-	-	<=0.343	-	35.25%	0.68	High
	-	>2.45	-	>0.343	-	7.55%	0.71	Low
	>2.50	<=2.45	-	>0.343	-	18.29%	0.94	Low

Crucial decision rules by using ID3 model_

Network type	Conditional factor					Criteria		Decision factor
	CB	CI	CC	CL	BC	Support	Confidence	PE
Dyadic network	-	<=25.5	-	-	-	62.86%	0.66	High
	-	>25.5	-	-	-	4.29%	1.00	Low
Star network	-		<=0.001	<=0.371		48.05%	0.72	High
	-	-	-	>0.371	-	7.03%	0.85	Low
Ringlike network	-	<=2.59	-	<=0.371	-	4.02%	0.66	Low
	-	(2.59,13.65]	-	<=0.371	-	9.12%	0.72	High
	-	>13.65	-	<=0.371	-	3.59%	1.00	Low
	-	-	-	>0.371	-	60.35%	0.90	Low
Complex network	-	-	-	<=0.339	-	44.37%	0.68	High
	-	>2.45	-	>0.339	-	7.34%	0.72	Low
	>2.50	<=2.45	-	>0.339	-	17.22%	0.95	Low

Basic network indicators of heterogeneous alliance network embedding types_

Type	Communities	Nodes	Average cluster coefficient	Average path length	Density	Average degree	Maximum diameter
Dyadic network	140	280	0.000	1.000	0.004	0.500	1.000
Star network	120	553	0.238	1.784	0.003	0.855	5.000
Ringlike network	60	309	0.875	1.560	0.011	1.618	3.000
Complex network	16	1,017	0.636	6.116	0.004	1.986	15.000

Crucial decision rules for various alliance network types_

Network type	Conditional factor					Criteria		Decision factor
	CB	CI	CC	CL	BC	Support	Confidence	PE
Dyadic network	-	<=25.5	-	-	-	62.86%	0.66	High
	-	>25.5	-	-	-	4.29%	1.00	Low
Star network	-	-	<=0.002	<=0.367		45.39%	0.70	High
	-	-	-	>0.367	-	7.23%	0.80	Low
Ringlike network	-	<=2.583	-	<=0.367	-	4.21%	0.65	Low
	-	(2.583,14]	-	<=0.367	-	8.74%	0.77	High
	-	>14	-	<=0.367	-	3.88%	1.00	Low
	-	-	-	>0.367	-	69.57%	0.89	Low
Complex network	-	-	-	<=0.339	-	38.54%	0.69	High
	-	>2.45	-	>0.339	-	8.85%	0.67	Low
	>2.50	<=2.45	-	>0.339	-	20.16%	0.97	Low