compactCreditScorecard Object Workflow
This example shows a workflow for creating a compactCreditScorecard object from a creditscorecard object.
Step 1. Create a creditscorecard object
To create a compactCreditScorecard object, you must first create a creditscorecard object. Create a creditscorecard object with the CreditCardData.mat file, and set the name-value pair argument 'BinMissingData' to true because the dataMissing data set contains missing data.
load CreditCardData.mat sc = creditscorecard(dataMissing,'IDVar','CustID','BinMissingData',true); sc = autobinning(sc); sc = modifybins(sc,'CustAge','MinValue',0); sc = modifybins(sc,'CustIncome','MinValue',0);
Step 2. Fit a logistic regression model for the creditscorecard object
Use fitmodel to fit a logistic regression model using the Weight of Evidence (WOE) data.
[sc, mdl] = fitmodel(sc);
1. Adding CustIncome, Deviance = 1490.8527, Chi2Stat = 32.588614, PValue = 1.1387992e-08
2. Adding TmWBank, Deviance = 1467.1415, Chi2Stat = 23.711203, PValue = 1.1192909e-06
3. Adding AMBalance, Deviance = 1455.5715, Chi2Stat = 11.569967, PValue = 0.00067025601
4. Adding EmpStatus, Deviance = 1447.3451, Chi2Stat = 8.2264038, PValue = 0.0041285257
5. Adding CustAge, Deviance = 1442.8477, Chi2Stat = 4.4974731, PValue = 0.033944979
6. Adding ResStatus, Deviance = 1438.9783, Chi2Stat = 3.86941, PValue = 0.049173805
7. Adding OtherCC, Deviance = 1434.9751, Chi2Stat = 4.0031966, PValue = 0.045414057
Generalized linear regression model:
logit(status) ~ 1 + CustAge + ResStatus + EmpStatus + CustIncome + TmWBank + OtherCC + AMBalance
Distribution = Binomial
Estimated Coefficients:
Estimate SE tStat pValue
________ ________ ______ __________
(Intercept) 0.70229 0.063959 10.98 4.7498e-28
CustAge 0.57421 0.25708 2.2335 0.025513
ResStatus 1.3629 0.66952 2.0356 0.04179
EmpStatus 0.88373 0.2929 3.0172 0.002551
CustIncome 0.73535 0.2159 3.406 0.00065929
TmWBank 1.1065 0.23267 4.7556 1.9783e-06
OtherCC 1.0648 0.52826 2.0156 0.043841
AMBalance 1.0446 0.32197 3.2443 0.0011775
1200 observations, 1192 error degrees of freedom
Dispersion: 1
Chi^2-statistic vs. constant model: 88.5, p-value = 2.55e-16
Step 3. Create a new data set for scoring the creditscorecard object
Create a new data set that is used for scoring based on the previously created creditscorecard object.
tdata = data(1:10, mdl.PredictorNames); tdata.CustAge(2) = NaN; tdata.CustAge(5) = -5; tdata.ResStatus(1) = missing; tdata.ResStatus(3) = 'Landlord'; tdata.EmpStatus(3) = missing; tdata.CustIncome(4) = NaN; tdata.EmpStatus(7) = 'Freelancer'; tdata.CustIncome(8) = -1; tdata.CustIncome(4) = NaN; disp(tdata);
CustAge ResStatus EmpStatus CustIncome TmWBank OtherCC AMBalance
_______ ___________ ___________ __________ _______ _______ _________
53 <undefined> Unknown 50000 55 Yes 1055.9
NaN Home Owner Employed 52000 25 Yes 1161.6
47 Landlord <undefined> 37000 61 No 877.23
50 Home Owner Employed NaN 20 Yes 157.37
-5 Home Owner Employed 53000 14 Yes 561.84
65 Home Owner Employed 48000 59 Yes 968.18
34 Home Owner Freelancer 32000 26 Yes 717.82
50 Other Employed -1 33 No 3041.2
50 Tenant Unknown 52000 25 Yes 115.56
49 Home Owner Unknown 53000 23 Yes 718.5
Use displaypoints to display the points per predictor. Use score to compute the credit scores using the new data (tdata). Then use probdefault with the new data (tdata) to calculate probability of default. When using formatpoints, the 'Missing' name-value pair argument is set to 'minpoints' because tdata contains missing data.
PointsInfo = displaypoints(sc)
PointsInfo=38×3 table
Predictors Bin Points
_____________ ______________ _________
{'CustAge' } {'[0,33)' } -0.14173
{'CustAge' } {'[33,37)' } -0.11095
{'CustAge' } {'[37,40)' } -0.059244
{'CustAge' } {'[40,46)' } 0.074167
{'CustAge' } {'[46,48)' } 0.1889
{'CustAge' } {'[48,51)' } 0.20204
{'CustAge' } {'[51,58)' } 0.22935
{'CustAge' } {'[58,Inf]' } 0.45019
{'CustAge' } {'<missing>' } 0.0096749
{'ResStatus'} {'Tenant' } -0.029778
{'ResStatus'} {'Home Owner'} 0.12425
{'ResStatus'} {'Other' } 0.36796
{'ResStatus'} {'<missing>' } 0.1364
{'EmpStatus'} {'Unknown' } -0.075948
{'EmpStatus'} {'Employed' } 0.31401
{'EmpStatus'} {'<missing>' } NaN
⋮
[Scores, Points] = score(sc, tdata)
Scores = 10×1
1.2784
1.0071
NaN
NaN
0.9960
1.8771
NaN
NaN
1.0283
0.8095
Points=10×7 table
CustAge ResStatus EmpStatus CustIncome TmWBank OtherCC AMBalance
_________ _________ _________ __________ _________ ________ _________
0.22935 0.1364 -0.075948 0.45309 0.3958 0.15715 -0.017438
0.0096749 0.12425 0.31401 0.45309 -0.033652 0.15715 -0.017438
0.1889 0.1364 NaN 0.080697 0.3958 -0.18537 -0.017438
0.20204 0.12425 0.31401 NaN -0.044701 0.15715 0.35539
0.0096749 0.12425 0.31401 0.45309 -0.044701 0.15715 -0.017438
0.45019 0.12425 0.31401 0.45309 0.3958 0.15715 -0.017438
-0.11095 0.12425 NaN -0.11452 -0.033652 0.15715 -0.017438
0.20204 0.36796 0.31401 NaN -0.033652 -0.18537 -0.21195
0.20204 -0.029778 -0.075948 0.45309 -0.033652 0.15715 0.35539
0.20204 0.12425 -0.075948 0.45309 -0.033652 0.15715 -0.017438
pd = probdefault(sc, tdata)
pd = 10×1
0.2178
0.2676
NaN
NaN
0.2697
0.1327
NaN
NaN
0.2634
0.3080
sc = formatpoints(sc,'BasePoints',true,'Missing','minpoints','Round','finalscore','PointsOddsAndPDO',[500, 2, 50]); PointsInfo1 = displaypoints(sc)
PointsInfo1=39×3 table
Predictors Bin Points
______________ ______________ _______
{'BasePoints'} {'BasePoints'} 500.66
{'CustAge' } {'[0,33)' } -17.461
{'CustAge' } {'[33,37)' } -15.24
{'CustAge' } {'[37,40)' } -11.511
{'CustAge' } {'[40,46)' } -1.8871
{'CustAge' } {'[46,48)' } 6.3888
{'CustAge' } {'[48,51)' } 7.3367
{'CustAge' } {'[51,58)' } 9.3068
{'CustAge' } {'[58,Inf]' } 25.238
{'CustAge' } {'<missing>' } -6.5392
{'ResStatus' } {'Tenant' } -9.3852
{'ResStatus' } {'Home Owner'} 1.7253
{'ResStatus' } {'Other' } 19.305
{'ResStatus' } {'<missing>' } 2.6022
{'EmpStatus' } {'Unknown' } -12.716
{'EmpStatus' } {'Employed' } 15.414
⋮
[Scores1, Points1] = score(sc, tdata)
Scores1 = 10×1
542
523
488
495
522
585
445
448
524
508
Points1=10×8 table
BasePoints CustAge ResStatus EmpStatus CustIncome TmWBank OtherCC AMBalance
__________ _______ _________ _________ __________ _______ _______ _________
500.66 9.3068 2.6022 -12.716 25.446 21.314 4.0988 -8.495
500.66 -6.5392 1.7253 15.414 25.446 -9.6646 4.0988 -8.495
500.66 6.3888 2.6022 -12.716 -1.4161 21.314 -20.609 -8.495
500.66 7.3367 1.7253 15.414 -42.148 -10.462 4.0988 18.399
500.66 -6.5392 1.7253 15.414 25.446 -10.462 4.0988 -8.495
500.66 25.238 1.7253 15.414 25.446 21.314 4.0988 -8.495
500.66 -15.24 1.7253 -12.716 -15.498 -9.6646 4.0988 -8.495
500.66 7.3367 19.305 15.414 -42.148 -9.6646 -20.609 -22.526
500.66 7.3367 -9.3852 -12.716 25.446 -9.6646 4.0988 18.399
500.66 7.3367 1.7253 -12.716 25.446 -9.6646 4.0988 -8.495
pd1 = probdefault(sc, tdata)
pd1 = 10×1
0.2178
0.2676
0.3721
0.3488
0.2697
0.1327
0.5178
0.5077
0.2634
0.3080
Step 4. Create a compactCreditScorecard object from the creditscorecard object
Create a compactCreditScorecard object using the creditscorecard object as the input. Alternatively, you can create the compactCreditScorecard object using the compact function in Financial Toolbox™.
csc = compactCreditScorecard(sc)
csc =
compactCreditScorecard with properties:
Description: ''
GoodLabel: 0
ResponseVar: 'status'
WeightsVar: ''
NumericPredictors: {'CustAge' 'CustIncome' 'TmWBank' 'AMBalance'}
CategoricalPredictors: {'ResStatus' 'EmpStatus' 'OtherCC'}
PredictorVars: {'CustAge' 'ResStatus' 'EmpStatus' 'CustIncome' 'TmWBank' 'OtherCC' 'AMBalance'}
Step 5. Use associated functions to analyze the compactCreditScorecard object
You can analyze the compactCreditScorecard object with displaypoints, score, and probdefault from Risk Management Toolbox™.
PointsInfo2 = displaypoints(csc)
PointsInfo2=39×3 table
Predictors Bin Points
______________ ______________ _______
{'BasePoints'} {'BasePoints'} 500.66
{'CustAge' } {'[0,33)' } -17.461
{'CustAge' } {'[33,37)' } -15.24
{'CustAge' } {'[37,40)' } -11.511
{'CustAge' } {'[40,46)' } -1.8871
{'CustAge' } {'[46,48)' } 6.3888
{'CustAge' } {'[48,51)' } 7.3367
{'CustAge' } {'[51,58)' } 9.3068
{'CustAge' } {'[58,Inf]' } 25.238
{'CustAge' } {'<missing>' } -6.5392
{'ResStatus' } {'Tenant' } -9.3852
{'ResStatus' } {'Home Owner'} 1.7253
{'ResStatus' } {'Other' } 19.305
{'ResStatus' } {'<missing>' } 2.6022
{'EmpStatus' } {'Unknown' } -12.716
{'EmpStatus' } {'Employed' } 15.414
⋮
[Scores2, Points2] = score(csc, tdata)
Scores2 = 10×1
542
523
488
495
522
585
445
448
524
508
Points2=10×8 table
BasePoints CustAge ResStatus EmpStatus CustIncome TmWBank OtherCC AMBalance
__________ _______ _________ _________ __________ _______ _______ _________
500.66 9.3068 2.6022 -12.716 25.446 21.314 4.0988 -8.495
500.66 -6.5392 1.7253 15.414 25.446 -9.6646 4.0988 -8.495
500.66 6.3888 2.6022 -12.716 -1.4161 21.314 -20.609 -8.495
500.66 7.3367 1.7253 15.414 -42.148 -10.462 4.0988 18.399
500.66 -6.5392 1.7253 15.414 25.446 -10.462 4.0988 -8.495
500.66 25.238 1.7253 15.414 25.446 21.314 4.0988 -8.495
500.66 -15.24 1.7253 -12.716 -15.498 -9.6646 4.0988 -8.495
500.66 7.3367 19.305 15.414 -42.148 -9.6646 -20.609 -22.526
500.66 7.3367 -9.3852 -12.716 25.446 -9.6646 4.0988 18.399
500.66 7.3367 1.7253 -12.716 25.446 -9.6646 4.0988 -8.495
pd2 = probdefault(csc, tdata)
pd2 = 10×1
0.2178
0.2676
0.3721
0.3488
0.2697
0.1327
0.5178
0.5077
0.2634
0.3080
Compare the size of the creditscorecard and compactCreditScorecard objects.
whos('dataMissing','sc','csc')
Name Size Bytes Class Attributes csc 1x1 41509 compactCreditScorecard dataMissing 1200x11 85035 table sc 1x1 167007 creditscorecard
The size of the compactCreditScorecard object is lightweight compared to the creditscorecard object. However, the compactCreditScorecard object cannot be directly modified. If you need to change a compactCreditScorecard object, you must change the starting creditscorecard object, and then reconvert that object to create the compactCreditScorecard object again.
See Also
creditscorecard | screenpredictors | autobinning | bininfo | predictorinfo | modifypredictor | modifybins | bindata | plotbins | fitmodel | displaypoints | formatpoints | score | setmodel | probdefault | validatemodel
Topics
- Common Binning Explorer Tasks
- Credit Scorecard Modeling with Missing Values
- Feature Screening with screenpredictors
- Troubleshooting Credit Scorecard Results
- Credit Rating by Bagging Decision Trees
- Stress Testing of Consumer Credit Default Probabilities Using Panel Data
- Overview of Binning Explorer
- About Credit Scorecards
- Credit Scorecard Modeling Workflow
- Monotone Adjacent Pooling Algorithm (MAPA)
- Credit Scorecard Modeling Using Observation Weights
