Business Process Modeling and Business Performance in the Mobile Industry
Background
Before starting the analysis process, it is necessary to give a brief description of the dependent variable and how it will be employed in the analysis in later stages. Mobile subscriber's retention behaviour as a dependent variable is used to check whether they have reached a predetermined decision to renew with or switch their mobile operators. The sample participants? possibility of repeat purchasing is used as a dependent variable that needs to be studied with respect to the two pre-behaviour and four post-behaviour predictors that affect subscribers in the retention situation. According to actual customer behaviour, current customer-supplier relationship, and mutual interaction, the repeat purchase question aims to check whether a customer is planning to renew his or her mobile service subscription or contract with his/her current mobile supplier or not. Based on different contextual and relational factors, a customer reaches a decision to renew with or switch his/her current mobile supplier when he or she compares, within the retention circumstances, what the current supplier is providing with what other mobile suppliers are offering in price plans containing a variety of reinforcements and punishments.
Data Analysis and Discussion
Analysis of the data is consisted of two parts. The first is the extraction of principal components for principle components analysis. For this purpose the correlation matrix is first obtained for the original variables. Then the number of principal components is determined using Eigen values and communalities. Successfully extracted principal components are then used to reproduce the correlation matrix. In the second part, dimensionality is tested and canonical correlation analysis is performed because the associations might have multiple dimensions.
The following table gives the correlation between the original variables. Obtaining the correlation matrix of original variables is appropriate prior to the principle component analysis. One reason is that very highly correlated variables needed to be removed from the analysis, by very high here it means above 0.9. When two variables have this much high correlation, the two variables seem to be measuring same thing. Such variables can be combined by various methods like taking average. For very low values of correlation between two or more variables, here by low value it means less than 0.1, these variables load only onto one principle component. In simpler words, variables with very low value of correlation make their own principle component. This is not appropriate as principle component analysis aims at reducing the number of variables. The following table depicts that no pair of variables have correlation above 0.9 hence each variable is supposed to be measuring different aspects. The variables for which value of correlation coefficient is less than 0.1 are not to be included in a component and are incorporated in the model as separate variables.
Table 1: Correlation Matrix of Original Variables
SQ1
SQ2
SQ3
SQ4
SQ5
SQ6
SQ7
CS1
CS2
CS3
CS4
BPC1
BPC2
BPC3
BPC4
BPT1
BPT2
BPT3
BPM1
BPM2
BPM3
BPM4
BPM5
BPM6
BPM7
SQ1
1.000
.839
.682
.704
.648
.081
.487
.551
.530
.555
.525
.405
.355
.361
.434
.474
.433
.486
.383
.219
.351
.353
.470
.338
.375
SQ2
.839
1.000
.756
.706
.625
.153
.449
.532
.511
.474
.485
.356
.316
.319
.343
.431
.403
.477
.388
.207
.401
.403
.489
.365
.442
SQ3
.682
.756
1.000
.702
.616
.239
.441
.498
.457
.460
.419
.314
.231
.281
.369
.347
.318
.446
.403
.236
.394
.396
.437
.292
.386
SQ4
.704
.706
.702
1.000
.787
.242
.557
.513
.535
.508
.445
.374
.279
.343
.453
.371
.325
.391
.411
.189
.421
.412
.469
.299
.364
SQ5
.648
.625
.616
.787
1.000
.345
.681
.597
.595
.591
.507
.411
.345
.385
.408
.369
.338
.400
.478
.251
.488
.499
.530
.282
.313
SQ6
.081
.153
.239
.242
.345
1.000
.412
.248
.199
.201
.100
.001
.049
.154
.049
.044
.025
.031
.132
.193
.182
.205
.245
.053
.211
SQ7
.487
.449
.441
.557
.681
.412
1.000
.615
.593
.483
.511
.452
.395
.480
.446
.490
.454
.428
.438
.278
.398
.389
.479
.200
.326
CS1
.551
.532
.498
.513
.597
.248
.615
1.000
.899
.706
.630
.424
.398
.372
.444
.509
.467
.433
.416
.179
.542
.553
.502
.117
.281
CS2
.530
.511
.457
.535
.595
.199
.593
.899
1.000
.713
.617
.442
.414
.364
.452
.519
.466
.391
.444
.183
.563
.562
.532
.109
.239
CS3
.555
.474
.460
.508
.591
.201
.483
.706
.713
1.000
.679
.387
.395
.339
.483
.511
.427
.353
.502
.269
.501
.501
.475
.195
.379
CS4
.525
.485
.419
.445
.507
.100
.511
.630
.617
.679
1.000
.312
.265
.342
.327
.446
.417
.419
.391
.292
.496
.497
.502
.149
.364
BPC1
.405
.356
.314
.374
.411
.001
.452
.424
.442
.387
.312
1.000
.881
.696
.610
.485
.484
.350
.360
.213
.332
.319
.413
.168
.254
BPC2
.355
.316
.231
.279
.345
.049
.395
.398
.414
.395
.265
.881
1.000
.744
.641
.461
.441
.302
.345
.203
.342
.329
.396
.120
.221
BPC3
.361
.319
.281
.343
.385
.154
.480
.372
.364
.339
.342
.696
.744
1.000
.685
.435
.433
.340
.348
.236
.326
.328
.431
.062
.241
BPC4
.434
.343
.369
.453
.408
.049
.446
.444
.452
.483
.327
.610
.641
.685
1.000
.520
.543
.387
.435
.241
.412
.389
.456
.164
.268
BPT1
.474
.431
.347
.371
.369
.044
.490
.509
.519
.511
.446
.485
.461
.435
.520
1.000
.868
.630
.474
.289
.410
.398
.513
.262
.406
BPT2
.433
.403
.318
.325
.338
.025
.454
.467
.466
.427
.417
.484
.441
.433
.543
.868
1.000
.687
.440
.245
.387
.376
.509
.224
.399
BPT3
.486
.477
.446
.391
.400
.031
.428
.433
.391
.353
.419
.350
.302
.340
.387
.630
.687
1.000
.412
.297
.303
.304
.400
.289
.436
BPM1
.383
.388
.403
.411
.478
.132
.438
.416
.444
.502
.391
.360
.345
.348
.435
.474
.440
.412
1.000
.525
.454
.448
.524
.225
.438
BPM2
.219
.207
.236
.189
.251
.193
.278
.179
.183
.269
.292
.213
.203
.236
.241
.289
.245
.297
.525
1.000
.335
.347
.422
.301
.382
BPM3
.351
.401
.394
.421
.488
.182
.398
.542
.563
.501
.496
.332
.342
.326
.412
.410
.387
.303
.454
.335
1.000
.993
.731
-.001
.281
BPM4
.353
.403
.396
.412
.499
.205
.389
.553
.562
.501
.497
.319
.329
.328
.389
.398
.376
.304
.448
.347
.993
1.000
.738
.002
.286
BPM5
.470
.489
.437
.469
.530
.245
.479
.502
.532
.475
.502
.413
.396
.431
.456
.513
.509
.400
.524
.422
.731
.738
1.000
.247
.353
BPM6
.338
.365
.292
.299
.282
.053
.200
.117
.109
.195
.149
.168
.120
.062
.164
.262
.224
.289
.225
.301
-.001
.002
.247
1.000
.491
BPM7
.375
.442
.386
.364
.313
.211
.326
.281
.239
.379
.364
.254
.221
.241
.268
.406
.399
.436
.438
.382
.281
.286
.353
.491
1.000
Table 2: KMO and Bartlett's Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy.
.879
Bartlett's Test of Sphericity
Approx. Chi-Square
2925.662
df
300
Sig.
.000
In the above table, value Kaiser-Meyer-Olkin Measure of Sampling Adequacy varies between 0 and ...