Appendix A – Notations of the Section 2.

A – Acquisition cost per customer

A' – Average acquisition cost per customer
At'' – Variable cost of acquisition the customer
At''' – fixed cost of acquisition the customer
a – Acquisition rate, given a specific level of acquisition costs A
a' – Probability the firm convinces a prospect to become a customer
Bt  – Potential benefit from the customer at period t
C – Annual costs
C' – Costs per period
C jt – Cost of sales per transaction j in period t
Ct – Total cost of generating the revenue Rt in period t
'
C t – Costs from the total of customers in period t
C t'' – Direct cost of servicing the customer at time t
C t''' – Cost of sales in period t
C t
' '' '
– Unit cost of goods sold to the customer in year t
C t
' '' ''
– Variable cost per unit sold to the customer in year t
Ct'''''' – Fixed cost associated with the customer in year t
c – Promotion costs per customer per year
c' – Retention cost per customer per year
c' ' – Mailing costs
c' ' ' – Average marketing costs per customer
ck t – Variable marketing cost, k, in period t
c m, t – Unit marketing cost for the customer in channel m in year t
cs – Promotion costs per customer per sales cycle
ct – Retention costs per customer in period t
ct' – Retention spending between t and t+1 (present value at time t)
ct'' – Mailing cost in month t for the customer
ct''' – Termination costs for the relationship with the customer
ct'''' – Variable costs of loyalty programs for the customer in year t
ct''''' – Fixed costs of loyalty programs for the customer in year t
CF V
– Vector of customer cash flows in any future period
CLV T – Column vector of expected present value over T periods
CLVr ' – Value of a customer with recency r’
CM NRT – Average contribution margin of base transaction
CM t – Average contribution margin in month t
CM Up – Average contribution margin of upgrading transactions
CM y – Contribution margin of the customer in transaction y

187
CoopVt – Cooperation value of the customer in period t
CPt – Customer potential = f (predicted sales volume, predicted profit, …)
CQt – Customer quality = f (sales per period, profit contributions, number of different
products, …)
CSt – Customer share = f SQt , SPt 
D – Discount factor (the inverse of one plus the discount rate)
D(t ) – Discount function
d – Defection rate
f – Predicted purchase frequency for the customer
g – Growth rate
GC – Yearly gross contribution margin per customer
GC ' – Average gross contribution
GCs – Gross contribution margin per customer per sales cycle
GCt – Gross contribution of the customer in month t
GC 't – Gross contributions from reference activities of the customer in period t
I – Identity matrix
i – Discount rate of the period
InfoVt – Information value of the customer in period t
InnoVt – Innovation value of the customer in period t
m – Margin or profit from a customer per period
m' – Margin from a customer per year
m' ' – Average gross margin per period
m' ' ' – Margin per transaction
md t – Margin of the direct transactions of the customer in period t
mt – Margin or contribution of each customer in period t
mWoM – Margin generated by referral behaviour of the customer in period t
m0 – Margin at time 0, or the minimum margin
m – Margin at infinite time period, or the maximum possible margin
n' – Number of transactions per year
n' ' – Number of years between two consecutive sales
n' ' ' ' – Number of marketing actions per year
n0 – Number of customers acquired in base year 0
nb – Initial customer base at the time of the determination of CLV
nk – Number of customers in the kth cohort
nm , t – Number of contacts to the customer in channel m in year t
n r ',k – Number of transactions generated by a customer of recency r’ in period k
nt – Number of base (i.e., non-upgrade) transactions made by the customer in period t
n't – Number of “upgrade” transactions by the customer in period t
n ' 't – Number of units purchased by the customer in year t
nt''' – Number of hours that the salesperson spends attempting to acquire the customer
in year t
nt  j – Number of customers in period t-j

188
pl – Response probability of a segment of recency l
Pu  j – Probability of purchase for the current year u
qk – Cumulated probability for a customer to be eliminated (purged) after k periods
R – Annual revenues
R' – Revenues per period
Rjt – Revenue per transaction j in period t
Rt – Revenue from the customer in period t
Rt' – Revenue from the total of customers in period t
R t
''
– Revenue per unit charged to the customer in year t
R t
A
– Autonomous revenue of the customer in period t
RtCS – Cross selling revenue of the customer in period t
RtUS – Up selling revenue of the customer in period t
r – Yearly retention rate
r' – Recency
r' ' – Recency limit beyond which a customer is ticked off the database
rc – Retention probability as a function of the retention budget
rj – Probability of customer retention in period j
rt – Probability of customer retention in period t
rs – Retention rate per sales cycle
r0 – Retention at time 0 or the minimum retention
r – Retention at infinite time period or the maximum possible retention
S – Transition matrix
s – Rate of change of margin from the minimum to maximum
s' – Rate of change of retention from the minimum to maximum
S t  – Survival probability at time t (survival function)
t – Time period
T – Time horizon for estimating CLV
T' – Total service period of the customer
T '' – Expected service period of the customer
T ''' – Period over which the customer is assumed to remain active
u – Year u
v(t ) – Expected customer value at time t
v ' (t ) – Expected value generated by the customer at time t
W – Aggregated weighting for the discounted margins generated by customer the
customer as a result of his lead user, reference and option value potential
Wt ' – Hourly wage for a salesperson calling on the customer in year t
 – Length of the average repurchase cycle
 f t  – Future profit contribution of the customer at period t
 p t  – Past profit contribution of the customer at period t
 (u ) – Profit per customer in year u

189
190
 Appendix B – Analysis of the functional form of covariates (fixed-telephone)

Lowess smoother Lowess smoother

1.5

1
1
Martingale-like resid.

Martingale-like resid.

.5
.5

0
0
-.5

-.5
0 200 400 600 800 0 500 1000 1500 2000 2500
mean_overall_revenues mean_int_in_duration
bandwidth = .8 bandwidth = .8

Lowess smoother Lowess smoother

1

1
Martingale-like resid.

Martingale-like resid.
.5

.5
0

0
-.5

-.5

0 500 1000 0 5 10 15 20 25
current_debts mean_loc_out_value
bandwidth = .8 bandwidth = .8

Lowess smoother Lowess smoother

1

1
Martingale-like resid.

Martingale-like resid.
.5

.5
0

0
-.5

-.5

0 50 100 150 0 2000 4000 6000 8000

mean_int_out_value mean_loc_in_duration
bandwidth = .8 bandwidth = .8

191
 Lowess smoother Lowess smoother
1

1
Martingale-like resid.

Martingale-like resid.
.5

.5
0

0
-.5

-.5
0 5 10 15 0 10 20 30 40 50
mean_nat_out_value mean_other_value
bandwidth = .8 bandwidth = .8

Lowess smoother Lowess smoother

1

1
Martingale-like resid.

Martingale-like resid.
.5

.5
0

0
-.5

-.5

0 1000 2000 3000 0 500 1000 1500

mean_nat_in_duration mean_other_duration
bandwidth = .8 bandwidth = .8

Lowess smoother Lowess smoother

1

1
Martingale-like resid.

Martingale-like resid.
.5

.5
0

0
-.5

-.5

0 50 100 150 200 0 100 200 300 400 500

mean_mobile_value mean_quantity_calls_out
bandwidth = .8 bandwidth = .8

192
Appendix C – Graphs of Schoenfeld residuals (fixed-telephone)

Test of PH Assumption Test of PH Assumption

20

6
scaled Schoenfeld - mean_int_out_value
0

4
-20

2
-40

0
-60

-2
0 500 1000 1500 2000 0 500 1000 1500 2000
Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

.8
.5

scaled Schoenfeld - mean_int_in_duration

.6
.4
0

.2
-.5

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

2.5

20
scaled Schoenfeld - mean_loc_out_value
2

10
1.5

0
1

-10
.5
0

-20

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8
 Test of PH Assumption Test of PH Assumption

6
.06

scaled Schoenfeld - mean_mobile_value

4
.04

2
.02

0
0

-2
-.02

-4
0 500 1000 1500 2000 0 500 1000 1500 2000
Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

40

4
scaled Schoenfeld - mean_other_value
30

2
20

0
10

-2
0
-10

-4

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

.2
.2

scaled Schoenfeld - mean_other_duration

.15

0
.1
.05

-.2
0
-.05

-.4

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

194
 Test of PH Assumption Test of PH Assumption

40
3

scaled Schoenfeld - flat_plan_teleph_1

30
2

20
1

10
0

0
-10
-1

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

10

40
scaled Schoenfeld - flat_plan_teleph_2
5

20
0

0
-20
-5

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

40

20
scaled Schoenfeld - flat_plan_teleph_3
30

10
20

0
10

-10
0
-10

-20

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

195
 Test of PH Assumption Test of PH Assumption
60

300
scaled Schoenfeld - castelo_branco
40

200
20

100
0

0
-100
-20

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

400

40
300

20
scaled Schoenfeld - coimbra
200

0
100

-20
-40
0
-100

-60

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

600
40
20

400
scaled Schoenfeld - evora
0

200
-20

0
-40

-200
-60

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

196
 Test of PH Assumption Test of PH Assumption

20
40
20

scaled Schoenfeld - lisboa

scaled Schoenfeld - faro

0
0

-20
-20

-40
-40
-60

-60
0 500 1000 1500 2000 0 500 1000 1500 2000
Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

400

200
300

150
scaled Schoenfeld - portalegre
200

100
100

50
0
0
-100

-50

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

20
50

scaled Schoenfeld - porto

0
-20
0

-40
-50

-60

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

197
 Test of PH Assumption
40

Test of PH Assumption

200
20

150
scaled Schoenfeld - vila_real
0

100
-20

50
-40
-60

0
0 500 1000 1500 2000
Time

-50
bandwidth = .8

0 500 1000 1500 2000

Time
bandwidth = .8

Test of PH Assumption
20

Test of PH Assumption
300
0

200
scaled Schoenfeld - viseu
-20

100
-40

0
-60

0 500 1000 1500 2000

Time
-100

bandwidth = .8

0 500 1000 1500 2000

Time
bandwidth = .8

Test of PH Assumption
100

Test of PH Assumption
6
50

scaled Schoenfeld - gender

4
2
0

0
-50

-2

0 500 1000 1500 2000

Time
-4

bandwidth = .8
0 500 1000 1500 2000
Time
bandwidth = .8

198
 
Appendix D – Graphs of  ln  ln  Sˆ  t  against ln  t  (fixed-telephone)
10


10
-ln[-ln(Survival Probability)]
8

8
6

6
4

4
2

2
0

0
2 4 6 8 2 4 6 8
ln(analysis time) ln(analysis time)

portability = 0 portability = 1 flat_plan_teleph_2 = 0 flat_plan_teleph_2 = 1

10

10
-ln[-ln(Survival Probability)]
8

8
6

6
4

4
2

2
0

2 4 6 8 2 4 6 8
ln(analysis time) ln(analysis time)

payment_method = 0 payment_method = 1 flat_plan_teleph_3 = 0 flat_plan_teleph_3 = 1

10

10
-ln[-ln(Survival Probability)]
8

8
6

6
4

4
2

2
0

2 4 6 8 2 4 6 8
ln(analysis time) ln(analysis time)

flat_plan_teleph_1 = 0 flat_plan_teleph_1 = 1 equipment_renting = 0 equipment_renting = 1

10

10
8
6

-ln[-ln(Survival Probability)]

8
4
2

6
0

2 4 6 8

4
ln(analysis time)

province = 1/province = 17 province = 2/province = 18

2
province = 3 province = 5
province = 6 province = 7
province = 8 province = 9

0
province = 10 province = 11
province = 12 province = 13 2 4 6 8
ln(analysis time)
province = 14 province = 15
province = 16 gender = 0 gender = 1

200
 Appendix E - Analysis of the functional form of covariates (ADSL)

Lowess smoother Lowess smoother

1.5

1
1
Martingale-like resid.

Martingale-like resid.

.5
.5

0
0
-.5

-.5
0 200 400 600 800 0 500000 1000000 1500000 2000000
mean_overall_revenues mean_internet_traffic
bandwidth = .8 bandwidth = .8

Lowess smoother Lowess smoother

1

1
Martingale-like resid.

Martingale-like resid.
.5

.5
0

0
-.5

-.5

0 500 1000 0 5 10 15 20 25
current_debts mean_value_additional_traffic
bandwidth = .8 bandwidth = .8

201
202
Appendix F - Graphs of Schoenfeld residuals (ADSL)

Test of PH Assumption Test of PH Assumption

.0015
20

scaled Schoenfeld - mean_internet_traffic

0

.001
-20

.0005
-40

0
-60

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

scaled Schoenfeld - mean_value_additional_traffic
.4

20
.2

10
0

0
-.2

-10
-.4

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

2.5

40
scaled Schoenfeld - payment_method

30
2
1.5

20
1

10
.5

0
0

-10

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

203
 Test of PH Assumption Test of PH Assumption
60

300
scaled Schoenfeld - castelo_branco
40

200
20

100
0

0
-100
-20

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

400

40
300

20
scaled Schoenfeld - coimbra
200

0
100

-20
-40
0
-100

-60

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

600
40
20

400
scaled Schoenfeld - evora
0

200
-20

0
-40

-200
-60

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

204
 Test of PH Assumption Test of PH Assumption

20
40
20

scaled Schoenfeld - lisboa

scaled Schoenfeld - faro

0
0

-20
-20

-40
-40
-60

-60
0 500 1000 1500 2000 0 500 1000 1500 2000
Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

400

200
300

150
scaled Schoenfeld - portalegre
200

100
100

50
0

0
-100

-50

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

20
50

scaled Schoenfeld - porto

0
-20
0

-40
-50

-60

0 500 1000 1500 2000 0 500 1000 1500 2000

Time Time
bandwidth = .8 bandwidth = .8

205
 Test of PH Assumption
40

Test of PH Assumption

200
20

150
scaled Schoenfeld - vila_real
0

100
-20

50
-40

0
-60

0 500 1000 1500 2000

Time

-50
bandwidth = .8

0 500 1000 1500 2000

Time
bandwidth = .8

Test of PH Assumption
20

Test of PH Assumption
300
0

200
scaled Schoenfeld - viseu
-20

100
-40

0
-60

0 500 1000 1500 2000

Time
-100

bandwidth = .8

0 500 1000 1500 2000

Time
bandwidth = .8

Test of PH Assumption
100

Test of PH Assumption
4
50

scaled Schoenfeld - gender

2
0

0
-2
-50

0 500 1000 1500 2000

-4

Time
bandwidth = .8
0 500 1000 1500 2000
Time
bandwidth = .8

206
Appendix G - Graphs of  ln  ln  Sˆ  t  against ln  t  (ADSL)  

-ln[-ln(Survival Probability)]

10
10

8
6
4
2
0
5

0 2 4 6 8
ln(analysis time)

province = 1/province = 17 province = 2/province = 18

province = 3 province = 5
province = 6 province = 7
province = 8 province = 9
0

province = 10 province = 11
0 2 4 6 8 province = 12 province = 13
ln(analysis time)
province = 14 province = 15
payment_method = 0 payment_method = 1 province = 16

10
10

-ln[-ln(Survival Probability)]

8
6
5

4
2
0

0 2 4 6 8 0 2 4 6 8
ln(analysis time) ln(analysis time)

equipment_renting = 0 equipment_renting = 1 gender = 0 gender = 1

10
5
0

0 2 4 6 8
ln(analysis time)

flat_plan_adsl_1 = 0 flat_plan_adsl_1 = 1
208
 Appendix H - Analysis of the functional form of covariates (fixed-telephone) (with
satisfaction level)

Lowess smoother Lowess smoother

1

1
Martingale-like resid.

Martingale-like resid.
.5

.5
0

0
-.5

-.5
0 50 100 150 0 100 200 300 400 500
mean_overall_revenues mean_int_in_duration
bandwidth = .8 bandwidth = .8

Lowess smoother Lowess smoother

1

1
Martingale-like resid.

Martingale-like resid.
.5

.5
0

0
-.5

-.5

0 50 100 150 0 2 4 6 8
current_debts mean_loc_out_value
bandwidth = .8 bandwidth = .8

Lowess smoother Lowess smoother

1

1
Martingale-like resid.

Martingale-like resid.
.5

.5
0

0
-.5

-.5

0 10 20 30 40 50 0 500 1000 1500 2000

mean_int_out_value mean_loc_in_duration
bandwidth = .8 bandwidth = .8

209
 Lowess smoother
1

Lowess smoother

1
Martingale-like resid.

.5

Martingale-like resid.

.5
0

0
-.5

0 2 4 6 8
mean_nat_out_value

-.5
bandwidth = .8
0 5 10 15
mean_other_value
bandwidth = .8

Lowess smoother
1

Lowess smoother
1
Martingale-like resid.

.5

Martingale-like resid.

.5
0

0
-.5

0 200 400 600

mean_nat_in_duration
-.5

bandwidth = .8
0 50 100 150 200 250
mean_other_duration
bandwidth = .8

Lowess smoother
1

Lowess smoother
1
Martingale-like resid.

.5

Martingale-like resid.

.5
0

0
-.5

0 10 20 30 40 50
mean_mobile_value
-.5

bandwidth = .8
0 50 100 150 200
mean_quantity_calls_out
bandwidth = .8

Appendix I – Graphs of Schoenfeld residuals (fixed-telephone) (with satisfaction level)

210
 Test of PH Assumption

scaled Schoenfeld - mean_int_out_value

2
Test of PH Assumption
15

1
10

0
5

-1
0

-2
200 400 600 800 1000 1200
Time
-5

bandwidth = .8
200 400 600 800 1000 1200
Time
bandwidth = .8

Test of PH Assumption

.1
scaled Schoenfeld - mean_int_in_duration

Test of PH Assumption
.4

.05
.2

0
0
-.2

-.05

200 400 600 800 1000 1200

Time
-.4

bandwidth = .8
200 400 600 800 1000 1200
Time
bandwidth = .8

Test of PH Assumption
15
scaled Schoenfeld - mean_loc_out_value

Test of PH Assumption
.2

10
.1

5
0

0
-.1

-5

200 400 600 800 1000 1200

-.2

Time
bandwidth = .8
200 400 600 800 1000 1200
Time
bandwidth = .8
 Test of PH Assumption Test of PH Assumption

1.5
.02

scaled Schoenfeld - mean_mobile_value

1
.01

.5
0

0
-.01

-.5
-.02

-1
200 400 600 800 1000 1200 200 400 600 800 1000 1200
Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

20

3
scaled Schoenfeld - mean_other_value

2
10

1
0

0
-10

-1

200 400 600 800 1000 1200 200 400 600 800 1000 1200
Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

.03

.2
scaled Schoenfeld - mean_other_duration
.02

0
.01

-.2
0
-.01

-.4

200 400 600 800 1000 1200 200 400 600 800 1000 1200
Time Time
bandwidth = .8 bandwidth = .8

212
 .5 Test of PH Assumption

Test of PH Assumption

4
0

scaled Schoenfeld - gender

2
0
-.5

-2
-1

-4
200 400 600 800 1000 1200
Time
bandwidth = .8
200 400 600 800 1000 1200
Time
bandwidth = .8

Test of PH Assumption
10

Test of PH Assumption
.5
5

scaled Schoenfeld - satisfaction

0
0

-.5
-5

-1
-10

200 400 600 800 1000 1200

Time
-1.5

bandwidth = .8
200 400 600 800 1000 1200
Time
bandwidth = .8

Test of PH Assumption
5
0
-5
-10

200 400 600 800 1000 1200

Time
bandwidth = .8

213
214
Appendix J – Graphs of  ln  ln  Sˆ  t    against ln  t  (fixed-telephone) (with

satisfaction level)

6
6

-ln[-ln(Survival Probability)]

5
5

4
4

3
3

2
2

1
4 5 6 7 8 4 5 6 7 8
ln(analysis time) ln(analysis time)

portability = 0 portability = 1 gender = 0 gender = 1

6
5
4
3
2
1

4 5 6 7 8
ln(analysis time)

payment_method = 0 payment_method = 1

215
216
 Appendix K - Analysis of the functional form of covariates (ADSL) (with satisfaction
level)

Lowess smoother Lowess smoother

1

1
Martingale-like resid.

Martingale-like resid.
.5

.5
0

0
-.5

-.5
0 50 100 150 0 100000 200000 300000 400000 500000
mean_overall_revenues mean_internet_traffic
bandwidth = .8 bandwidth = .8

Lowess smoother Lowess smoother

1

1
Martingale-like resid.

Martingale-like resid.
.5

.5
0

0
-.5

-.5

0 50 100 150 0 2 4 6
current_debts mean_value_additional_traffic
bandwidth = .8 bandwidth = .8

217
218
Appendix L – Graphs of Schoenfeld residuals (ADSL) (with satisfaction level)

Test of PH Assumption Test of PH Assumption

15

.00002 .00004 .00006 .00008

scaled Schoenfeld - mean_internet_traffic
10
5
0

0
-.00002
-5

200 400 600 800 1000 1200 200 400 600 800 1000 1200
Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

scaled Schoenfeld - mean_value_additional_traffic
.15

6
.1

4
.05

2
0

0
-.05
-.1

-2

200 400 600 800 1000 1200 200 400 600 800 1000 1200
Time Time
bandwidth = .8 bandwidth = .8

Test of PH Assumption Test of PH Assumption

5
.1

scaled Schoenfeld - payment_method

0

0
-.1

-5
-.2

-10
-.3

200 400 600 800 1000 1200 200 400 600 800 1000 1200
Time Time
bandwidth = .8 bandwidth = .8

219
 Test of PH Assumption Test of PH Assumption
4

.5
scaled Schoenfeld - satisfaction
2

0
0

-.5
-2
-4

-1
200 400 600 800 1000 1200 200 400 600 800 1000 1200
Time Time
bandwidth = .8 bandwidth = .8

220
  
Appendix M – Graphs of  ln  ln  Sˆ  t  against ln  t  (ADSL) (with satisfaction level)

6
6

-ln[-ln(Survival Probability)]

5
5

4
4

3
3

2
2
1

1
4 5 6 7 8 4 5 6 7 8
ln(analysis time) ln(analysis time)

payment_method = 0 payment_method = 1 gender = 0 gender = 1

221