Impute Missing Data in the Credit Scorecard Workflow Using the Random Forest Algorithm
This example shows how to perform imputation of missing data in the credit scorecard workflow using the random forest algorithm.
Random forests are an ensemble learning method for classification or regression that operates by constructing a multitude of decision trees at training time and obtaining the class that is the mode of the classes (classification) or mean prediction (regression) of the individual trees. Random forests correct for the tendency of decision trees to overfit to the training set. For more information on the random forest algorithm, see fitrensemble and fitcensemble.
For additional information on alternative approaches for "treating" missing data, see Credit Scorecard Modeling with Missing Values.
Impute Missing Data Using Random Forest Algorithm
Use the dataMissing data set to impute missing values for the CustAge (numeric) and ResStatus (categorical) predictors.
load CreditCardData.mat
disp(head(dataMissing)); CustID CustAge TmAtAddress ResStatus EmpStatus CustIncome TmWBank OtherCC AMBalance UtilRate status
______ _______ ___________ ___________ _________ __________ _______ _______ _________ ________ ______
1 53 62 <undefined> Unknown 50000 55 Yes 1055.9 0.22 0
2 61 22 Home Owner Employed 52000 25 Yes 1161.6 0.24 0
3 47 30 Tenant Employed 37000 61 No 877.23 0.29 0
4 NaN 75 Home Owner Employed 53000 20 Yes 157.37 0.08 0
5 68 56 Home Owner Employed 53000 14 Yes 561.84 0.11 0
6 65 13 Home Owner Employed 48000 59 Yes 968.18 0.15 0
7 34 32 Home Owner Unknown 32000 26 Yes 717.82 0.02 1
8 50 57 Other Employed 51000 33 No 3041.2 0.13 0
Remove the 'CustID' and 'status' columns in the imputation process as these are the id and response values respectively. Alternatively, you can choose to leave the 'status' column in.
dataToImpute = dataMissing(:,setdiff(dataMissing.Properties.VariableNames,... {'CustID','status'},'stable')); rfImputedData = dataMissing;
Because multiple predictors contain missing data, turn on the 'Surrogate' flag when you create the decision tree template.
rng('default'); tmp = templateTree('Surrogate','on','Reproducible',true);
Next, use the fitrensemble and fitcensemble functions, which return the trained regression and classification ensemble model objects contain the results of boosting 100 regression and classification trees using LSBoost, respectively.
missingCustAge = ismissing(dataToImpute.CustAge); % Fit ensemble of regression learners rfCustAge = fitrensemble(dataToImpute,'CustAge','Method','Bag',... 'NumLearningCycles',200,'Learners',tmp,'CategoricalPredictors',... {'ResStatus','EmpStatus','OtherCC'}); rfImputedData.CustAge(missingCustAge) = predict(rfCustAge,... dataToImpute(missingCustAge,:)); missingResStatus = ismissing(dataToImpute.ResStatus); % Fit ensemble of classification learners rfResStatus = fitcensemble(dataToImpute,'ResStatus','Method','Bag',... 'NumLearningCycles',200,'Learners',tmp,'CategoricalPredictors',... {'EmpStatus','OtherCC'}); rfImputedData.ResStatus(missingResStatus) = predict(rfResStatus,... dataToImpute(missingResStatus,:)); % Optionally, round the age to the nearest integer rfImputedData.CustAge = round(rfImputedData.CustAge);
Compare Imputed Data to Original Data
disp(rfImputedData(5:10,:));
CustID CustAge TmAtAddress ResStatus EmpStatus CustIncome TmWBank OtherCC AMBalance UtilRate status
______ _______ ___________ __________ _________ __________ _______ _______ _________ ________ ______
5 68 56 Home Owner Employed 53000 14 Yes 561.84 0.11 0
6 65 13 Home Owner Employed 48000 59 Yes 968.18 0.15 0
7 34 32 Home Owner Unknown 32000 26 Yes 717.82 0.02 1
8 50 57 Other Employed 51000 33 No 3041.2 0.13 0
9 50 10 Tenant Unknown 52000 25 Yes 115.56 0.02 1
10 49 30 Home Owner Unknown 53000 23 Yes 718.5 0.17 1
disp(rfImputedData(find(missingCustAge,5),:));
CustID CustAge TmAtAddress ResStatus EmpStatus CustIncome TmWBank OtherCC AMBalance UtilRate status
______ _______ ___________ __________ _________ __________ _______ _______ _________ ________ ______
4 54 75 Home Owner Employed 53000 20 Yes 157.37 0.08 0
19 54 14 Home Owner Employed 51000 11 Yes 519.46 0.42 1
138 52 31 Other Employed 41000 2 Yes 1101.8 0.32 0
165 46 21 Home Owner Unknown 38000 70 No 1217 0.2 0
207 52 38 Home Owner Employed 48000 12 No 573.9 0.1 0
disp(rfImputedData(find(missingResStatus,5),:));
CustID CustAge TmAtAddress ResStatus EmpStatus CustIncome TmWBank OtherCC AMBalance UtilRate status
______ _______ ___________ __________ _________ __________ _______ _______ _________ ________ ______
1 53 62 Tenant Unknown 50000 55 Yes 1055.9 0.22 0
22 51 13 Tenant Employed 35000 33 Yes 468.85 0.01 0
33 46 8 Home Owner Unknown 32000 26 Yes 940.78 0.3 0
47 52 56 Tenant Employed 56000 79 Yes 294.46 0.12 0
103 64 49 Home Owner Employed 50000 35 Yes 118.43 0 0
Plot a histogram of the predictor values before and after imputation.
Predictor ="CustAge"; f1 = figure; ax1 = axes(f1); histogram(ax1,rfImputedData.(Predictor),'FaceColor','red','FaceAlpha',1); hold on histogram(ax1,dataMissing.(Predictor),'FaceColor','blue','FaceAlpha',1); legend(strcat("Imputed ", Predictor), strcat("Observed ", Predictor)); title(strcat("Histogram of ", Predictor));
Create Credit Scorecard Model Using New Imputed Data
Use the imputed data to create the creditscorecard object, and then use autobinning, fitmodel, and formatpoints to create a credit scorecard model.
sc = creditscorecard(rfImputedData,'IDVar','CustID'); sc = autobinning(sc); [sc,mdl] = fitmodel(sc,'display','off'); sc = formatpoints(sc,'PointsOddsAndPDO',[500 2 50]); PointsInfo = displaypoints(sc); disp(PointsInfo);
Predictors Bin Points
______________ _____________________ ______
{'CustAge' } {'[-Inf,33)' } 83.828
{'CustAge' } {'[33,37)' } 86.511
{'CustAge' } {'[37,40)' } 88.307
{'CustAge' } {'[40,46)' } 97.562
{'CustAge' } {'[46,48)' } 106.75
{'CustAge' } {'[48,51)' } 107.21
{'CustAge' } {'[51,58)' } 108.57
{'CustAge' } {'[58,Inf]' } 123.71
{'CustAge' } {'<missing>' } NaN
{'EmpStatus' } {'Unknown' } 87.502
{'EmpStatus' } {'Employed' } 115.45
{'EmpStatus' } {'<missing>' } NaN
{'CustIncome'} {'[-Inf,29000)' } 58.301
{'CustIncome'} {'[29000,33000)' } 84.752
{'CustIncome'} {'[33000,35000)' } 96.533
{'CustIncome'} {'[35000,40000)' } 98.73
{'CustIncome'} {'[40000,42000)' } 99.542
{'CustIncome'} {'[42000,47000)' } 110.97
{'CustIncome'} {'[47000,Inf]' } 125.39
{'CustIncome'} {'<missing>' } NaN
{'TmWBank' } {'[-Inf,12)' } 79.83
{'TmWBank' } {'[12,23)' } 89.717
{'TmWBank' } {'[23,45)' } 90.511
{'TmWBank' } {'[45,71)' } 121.36
{'TmWBank' } {'[71,Inf]' } 161.25
{'TmWBank' } {'<missing>' } NaN
{'AMBalance' } {'[-Inf,558.88)' } 118.25
{'AMBalance' } {'[558.88,1254.28)' } 91.772
{'AMBalance' } {'[1254.28,1597.44)'} 88.425
{'AMBalance' } {'[1597.44,Inf]' } 77.96
{'AMBalance' } {'<missing>' } NaN
Calculate Scores and Probability of Default for New Customers
Create a data set of 'new customers' and then calculate the scores and probabilities of default.
dataNewCustomers = dataMissing(1:20,1:end-1); disp(head(dataNewCustomers));
CustID CustAge TmAtAddress ResStatus EmpStatus CustIncome TmWBank OtherCC AMBalance UtilRate
______ _______ ___________ ___________ _________ __________ _______ _______ _________ ________
1 53 62 <undefined> Unknown 50000 55 Yes 1055.9 0.22
2 61 22 Home Owner Employed 52000 25 Yes 1161.6 0.24
3 47 30 Tenant Employed 37000 61 No 877.23 0.29
4 NaN 75 Home Owner Employed 53000 20 Yes 157.37 0.08
5 68 56 Home Owner Employed 53000 14 Yes 561.84 0.11
6 65 13 Home Owner Employed 48000 59 Yes 968.18 0.15
7 34 32 Home Owner Unknown 32000 26 Yes 717.82 0.02
8 50 57 Other Employed 51000 33 No 3041.2 0.13
Predict missing data in the scoring data set with the same imputation model as before.
missingCustAgeNewCustomers = isnan(dataNewCustomers.CustAge); missingResStatusNewCustomers = ismissing(dataNewCustomers.ResStatus); imputedCustAgeNewCustomers = round(predict(rfCustAge, dataNewCustomers(missingCustAgeNewCustomers,:))); imputedResStatusNewCustomers = predict(rfResStatus, dataNewCustomers(missingResStatusNewCustomers,:)); dataNewCustomers.CustAge(missingCustAgeNewCustomers) = imputedCustAgeNewCustomers; dataNewCustomers.ResStatus(missingResStatusNewCustomers) = imputedResStatusNewCustomers;
Use score to calculate the scores of the new customers.
[scores, points] = score(sc, dataNewCustomers); disp(scores);
534.5927 546.8314 534.0648 557.3729 546.0376 577.6807 441.0492 516.5229 528.8639 502.3882 501.3591 513.6481 534.6783 493.3650 541.1380 482.9742 482.5726 465.5446 547.4855 485.1795
disp(points);
CustAge EmpStatus CustIncome TmWBank AMBalance
_______ _________ __________ _______ _________
108.57 87.502 125.39 121.36 91.772
123.71 115.45 125.39 90.511 91.772
106.75 115.45 98.73 121.36 91.772
108.57 115.45 125.39 89.717 118.25
123.71 115.45 125.39 89.717 91.772
123.71 115.45 125.39 121.36 91.772
86.511 87.502 84.752 90.511 91.772
107.21 115.45 125.39 90.511 77.96
107.21 87.502 125.39 90.511 118.25
107.21 87.502 125.39 90.511 91.772
108.57 87.502 96.533 90.511 118.25
107.21 87.502 110.97 89.717 118.25
123.71 87.502 125.39 79.83 118.25
97.562 87.502 99.542 90.511 118.25
106.75 115.45 110.97 89.717 118.25
86.511 115.45 98.73 90.511 91.772
88.307 115.45 96.533 90.511 91.772
83.828 115.45 58.301 89.717 118.25
108.57 115.45 125.39 79.83 118.25
108.57 87.502 110.97 89.717 88.425