Choose Your Preset Values to Start

Press "Run Model" to Continue

Global Model

Below is our Input Panel. Click on variable names or check Appendix for more details. Model will take a few seconds to run.








NumTotalTrades(No Constraint)






PercentInstallTrades(No Constraint)


NumInstallTradesWBalance(No Constraint)





NumRevolvingTradesWBalance(No Constraint)


PercentTradesWBalance(No Constraint)

Original Features These are 23 original features given in the dataset. Subscale Features These are 10 composite features obtained from the previous 23 orginal features. Output This is the Risk Performance estimated from our model, which consists of only two outcomes: Good and Bad. Subscale Contribution Low Risk (No Default) High Risk (Default) ExternalRiskEstimate MSinceOldestTradeOpen MSinceMostRecentTradeOpen AverageMInFile NumSatisfactoryTrades NumTrades60Ever2DerogPubRec NumTrades90Ever2DerogPubRec NumTotalTrades NumTradesOpeninLast12M PercentTradesNeverDelq MSinceMostRecentDelq MaxDelq2PublicRecLast12M MaxDelqEver PercentInstallTrades NetFractionInstallBurden NumInstallTradesWBalance MSinceMostRecentInqexcl7days NumInqLast6M NumInqLast6Mexcl7days NetFractionRevolvingBurden NumRevolvingTradesWBalance NumBank2NatlTradesWHighUtilization PercentTradesWBalance ExternalRiskEstimate TradeOpenTime NumSatisfactoryTrades TradeFrequency Delinquency Installment Inquiry RevolvingBalance Utilization TradeWBalance Output

Local Models

Most Important Contributing Factors

This is a list of factors that contribute most heavily to the final prediction in our model.

Consistent Rule-based Explanations

The system solves an optimization problem (using Gurobi(c)) to compute the smallest set of rules that guarantees identical prediction by our global model.

Case-based Explanations

The table below compares the examined case with cases that share the same prediction outcome.


  • Highlighted in light blue are feature values of previous cases that are close to the examined case.
  • The emphasized features, highlighted in dark blue, are those used by the rules in the rule-based explanations.
  • More Information

    See documentation for more description of the global model and its local models (link is at the top of the page).


    Feature Explanations

    Variable Names Description Monotonicity Constraint (with respect to probability of bad=1) Role Subscale Features
    RiskPerformance Paid as negotiated flag (12-36 months). String of Good and Bad target
    ExternalRiskEstimate Consolidated version of risk markers Monotonically Decreasing predictor ExternalRiskEstimate
    MSinceOldestTradeOpen Months Since Oldest Trade Open Monotonically Decreasing predictor TradeOpenTime
    MSinceMostRecentTradeOpen Months Since Most Recent Trade Open Monotonically Decreasing predictor TradeOpenTime
    AverageMInFile Average Months in File Monotonically Decreasing predictor TradeOpenTime
    NumSatisfactoryTrades Number Satisfactory Trades Monotonically Decreasing predictor NumSatisfactoryTrades
    NumTrades60Ever2DerogPubRec Number Trades 60+ Ever Monotonically Increasing predictor TradeFrequency
    NumTrades90Ever2DerogPubRec Number Trades 90+ Ever Monotonically Increasing predictor TradeFrequency
    NumTotalTrades Number of Total Trades (total number of credit accounts) No constraint predictor TradeFrequency
    NumTradesOpeninLast12M Number of Trades Open in Last 12 Months Monotonically Increasing predictor TradeFrequency
    PercentTradesNeverDelq Percent Trades Never Delinquent Monotonically Decreasing predictor Delinquency
    MSinceMostRecentDelq Months Since Most Recent Delinquency Monotonically Decreasing predictor Delinquency
    MaxDelq2PublicRecLast12M Max Delq/Public Records Last 12 Months. See table "MaxDelq" for each category Values 0-7 are monotonically decreasing predictor Delinquency
    MaxDelqEver Max Delinquency Ever. See table "MaxDelq" for each category Values 2-8 are monotonically decreasing predictor Delinquency
    PercentInstallTrades Percent Installment Trades No constraint predictor Installment
    NetFractionInstallBurden Net Fraction Installment Burden. This is installment balance divided by original loan amount Monotonically Increasing predictor Installment
    NumInstallTradesWBalance Number Installment Trades with Balance No constraint predictor Installment
    MSinceMostRecentInqexcl7days Months Since Most Recent Inquiry excluding 7 days Monotonically Decreasing predictor Inquiry
    NumInqLast6M Number of Inquiries in Last 6 Months Monotonically Increasing predictor Inquiry
    NumInqLast6Mexcl7days Number of Inquiries in Last 6 Months excluding 7 days. Excluding the last 7 days removes inquiries that are likely due to price comparison shopping Monotonically Increasing predictor Inquiry
    NetFractionRevolvingBurden Net Fraction Revolving Burden. This is revolving balance divided by credit limit Monotonically Increasing predictor RevolvingBalance
    NumRevolvingTradesWBalance Number Revolving Trades with Balance No constraint predictor RevolvingBalance
    NumBank2NatlTradesWHighUtilization Number Bank/Natl Trades with high utilization ratio Monotonically Increasing predictor Utilization
    PercentTradesWBalance Percent Trades with Balance No constraint predictor TradeWBalance

    MaxDelq Table

    Value Meaning
    0 derogatory comment
    1 120+ days delinquent
    2 90 days delinquent
    3 60 days delinquent
    4 30 days delinquent
    5,6 unknown delinquent
    7 current and never delinquent
    8,9 all other
    Value Meaning
    1 No such value
    2 derogatory comment
    3 120+ days delinquent
    4 90 days delinquent
    5 60 days delinquent
    6 30 days delinquent
    7 unknown delinquency
    8 current and never deliquent
    9 all other

    Special Values

    Value Meaning
    -9 No Bureau Record or No Investigation
    -8 No Usable/Valid Trades or Inquiries
    -7 Condition not Met (e.g. No Inquiries, No Delinquencies)