Questions tagged [credit-scoring]
In finances, a credit score is a number representing the creditworthiness of a person.
68
questions
0
votes
0
answers
9
views
Credit scoring modeling [duplicate]
I am currently learning about credit scoring and anything relate to the field. I understand how logistic regression works to provide the credit score but i want to generate a score for each bin of ...
0
votes
0
answers
51
views
Can I use Information Value in a ML algorithm like random forest?
I have a logistic regression model used in credit scoring and I am studying the metrics used in the credit scoring field to evaluate the quality of the model. My ultimate objective is to understand if ...
0
votes
2
answers
77
views
Is probability of default an elusive concept?
Here, page 5, it is stated that probability of default is an elusive concept:
Thus, the realization of the random variable "default of firm n at time t" is governed by its PD if a firm has ...
- 13
0
votes
0
answers
37
views
How to predict partial events such as partial payments using machine learning or survival analysis?
I am trying to model and predict how long it takes for customers to make payments after they are billed, given a number of covariates/features.
At first sight, it rather looks like a survival analysis ...
- 1
0
votes
0
answers
88
views
Absorbing response in logit analysis of longitudinal data
In Applied Logistic Regression, chapter 9, logit analysis of correlated data is introduced by this example:
[...] consider a study of asthma in children in which subjects are interviewed
bi-monthly ...
- 13
0
votes
2
answers
66
views
Logistic regression for observations on a monthly grid - impact of overlapping information of a given customer
My question is based on the usual setting in credit scoring. Assume we have historical monthly observations of customers, their risk factors $R_{i,t}$ and the flag whether they are currently in ...
- 3,286
0
votes
0
answers
99
views
Binning and WoE transformation. Reducing number of categories for high cardinality features
I'm doing a credit default risk project. I have some features like a job title that has >100000 unique titles. What is the best way to reduce cardinality in a meaningful way?
The end goal is to get ...
0
votes
0
answers
329
views
Gini Coefficient and KS statistic conflict
What would it mean if the interpretation of the gini coefficient and KS statistic conflict with one another? In a scorecard model, if they were to conflict should you rely on one over the other?
- 1
1
vote
0
answers
29
views
Which Proper Scoring rule for when? [duplicate]
Hi, I'm quite new to statistics and have been tasked to evaluate if there is a difference in accuracy between 2 subpopulations in a logistic model.
The credit scoring company's model calculates the ...
- 11
18
votes
2
answers
821
views
Why is Binning, Weight of Evidence and Information Value so ubiquitous in the Credit Risk/Finance industry?
In the credit risk industry (and finance industry as a whole, at least here in the UK), there is a very common and accepted 'proper' way to build scorecards.
The general framework seems to be:
...
- 379
1
vote
1
answer
100
views
How to justify to create sub-models by slice of population?
I'm in charge of a scoring project and I'm getting average results with my binary classification model.
A large number of reasons could explain this, but I think one of them is the heterogeneity of ...
- 11
2
votes
1
answer
45
views
Single or multiple model. How to know beforehand?
I am building a probability of default model based on behavioral information. The dataset is a loan portfolio, which contains 4 types of loans: mortgage, unsecured loans, car loans and credit cards. ...
- 451
1
vote
0
answers
32
views
Resources on Weight of Evidence?
I've been toying (and getting some good results) with Weight of Evidence (and associated Information Value) to build logistic regression baseline for ML problems. If I remember correctly I've been ...
1
vote
2
answers
330
views
Interpretation of Logistic Regression output in Credit Scoring
Currently I'm working on a project involving loan data where I am trying to build a model that will predict a probability of an individual defaulting on a loan. I have binary dependent variable where ...
0
votes
1
answer
164
views
Callibration after oversampling
We have build a credit scoring model with OptBinning library in python.
In the process we oversampled the minority class and now we want to callibrate it back to ...
- 249
0
votes
1
answer
46
views
How does one handle sub categories while model building?
I'm building a classification model for predicting if someone would be a loan defaulter or not. Among the other 45 features, I have two features, they represent grade assigned by the bank to the ...
- 103
1
vote
1
answer
2k
views
Very high information value - what does it mean?
Hey I used iv function from scorecard package to calculate Information Value of my independent variables. What suprised me is ...
- 17
0
votes
1
answer
56
views
Streamlining / Optimizing conditional Rules in a rule engine using ML techniques
The problem is defined as follows :
We are dealing with a Rule engine used to classify a credit risk as 'Good', 'Medium' or 'bad'. Say the rule engine which has say 10 rules. These rules are ...
- 295
0
votes
0
answers
68
views
Trade-off between variable stability and discriminatory model quality
It is very normal for data scientists and modeling professionals to be concerned with the stability of the model. It basically means that if a variable is important today, it cannot change its ...
- 165
1
vote
0
answers
39
views
Credit risk book reference
Credit risk is a beautiful field that relies on basic notions of statistics and stochastic processes.
I have been studying it, and now I am trying to understand the market models such as KMV, ...
1
vote
2
answers
3k
views
How to convert coefficients of logistic regression into score values when constructing scorecard?
I have built a logistic regression model which takes a dataframe of dummy values as an input and produces binary classification (0 for accept, 1 for default). I have already split all variables into ...
- 132
0
votes
0
answers
54
views
4
votes
3
answers
156
views
How to prove variable is ordinal
I want to run a model that determines the probability of self curing for loan customers who entered arrears.
The thing is that when a customer enters arrears there are 3 ways he could go:
-not cure
-...
- 61
1
vote
0
answers
129
views
What happens with the significance of binned variables?
For this project I was required to create a credit risk scorecard witht the 4 most relevant variables, so I binned all variables and selected them by chi2 and IV.
I ran the logistic and linear ...
- 61
1
vote
1
answer
40
views
Getting scores from PCA?
I have performed PCA in my dataset. The first 3 components explain 90% variance of the data points. Can I add these principle components to get a score?
...
3
votes
1
answer
344
views
How does probability of default evolve over time?
Say I have a probability of default of 0.02 (which is annual so over next year) for a certain client. Then say this client takes out a 180 day loan, how can I adjust my probability of default for this ...
- 83
0
votes
0
answers
1k
views
Anchor Point Calibration (Cyclicality, Central Tendency) - Probability of Default (PD) Question about assessment - Credit Risk
Let's assume we have a PD model that was calibrated with the following parameters:
...
- 483
1
vote
1
answer
491
views
Target Score and Target Odds in Credit Scorecard development
I am working on building a Credit Scorecard model. Till now I have performed the following steps:
Data preparation and cleansing
Calculate WOE and IV values
Model Fitting
Now I working on ...
1
vote
0
answers
341
views
High cross validation score but low model performance on test set
I'm doing a machine learning project and need to predict a user's credit default probability. I tried some simple automated feature engineering and got a good AUC score on training set using ...
- 111
1
vote
0
answers
54
views
How to transform/convert likelihoods to scores?
I have the probability of loan default for a labeled dataset where the distribution of probabilities is heavily skewed. Labels are defined as "good/0" for no default and "bad/1" for defaults. My goal ...
- 11
1
vote
1
answer
95
views
To determine variables to figure out the bad customers in credit risk modeling [closed]
I am developing a probability to default model on a data from landing firm. After running the GLM() model i have got the below message:
...
- 153
0
votes
0
answers
114
views
Is the regression coefficient the same for all categories of a categorical variable?
Let's imagine I build a scorecard with a single binned variable that can only take two values. In the weight of reference framework I would replace the two possible values by their weight of evidence ...
- 101
2
votes
1
answer
208
views
German credit dataset : interpretation of checking_status feature
I am struggling to understand the meaning of some features of the german credit dataset.
I am particularly interested in, the categorical feature checking_status (Status of existing checking account) ...
- 121
1
vote
1
answer
141
views
What is the relationship between the risk ratio and the weight of evidence? [duplicate]
I've been reading about risk ratios as typical measures in clinical settings. In the finance and credit literature, there is the weight of evidence (WoE) measure that is used to encode and study ...
- 1,439
0
votes
1
answer
1k
views
Credit scorecard using Logistic Regression on R
I tried developing a scorecard to assess creditworthiness using the "Scorecard" package on R. The problem I encountered is when I scale the card and calculate the points for each attribute of each ...
- 139
1
vote
0
answers
286
views
Using categories instead of WoE values
As for as I can understand, the Weight of Evidence strategy is the following:
For continuous independent variables : First, create bins (categories / groups) for a continuous independent variable and ...
- 63
0
votes
0
answers
252
views
Credit Scoring WoE Calculation
I'm creating credit scoring model and stuck with WoE calculation. I know the formula and I know how to compute WoE for train sample. Should I use train sample WoE for test sample or I should compute ...
- 61
0
votes
0
answers
23
views
Using posterior variable in credit risk model
I am rebuilding a credit risk model using logistic regression (either ridge penalty or elasticnet) to predict first payment default. Historically, the company approves an applicant for a loan to ...
1
vote
0
answers
526
views
How do unsupervised credit scoring models that don't consider historical financial data work?
There seems to be a number of startups (Zest Finance, Credolab etc.) that provide credit scoring schemes that rely exclusively on alternative data without considering users historical financial data ...
- 111
0
votes
1
answer
69
views
Hypothesis testing two sided tail test
I have a bank customer loan dataset with columns loan amount, funded amount, interest rate(high, medium, low), annual income of customer, loan status as (default and fully paid).
Could I use two ...
Shivam Sinha
3
votes
1
answer
179
views
Behavioral credit scoring: problems
I would like to create a behavioral credit scoring model to score the applications for which transaction data is available.
There's an obvious problem mentioned in Thomas et al. Credit Scoring and Its ...
- 297
2
votes
0
answers
768
views
Using bayes theorem to calculate credit risk given prior knowledge and predicted probability
How can one combine:
a priori knowledge of the default proability of a certain loan type based on historical data
the default probability of an individual loan as predicted by a machine learning ...
- 33
1
vote
1
answer
385
views
Comparing coefficient across datasets - Cox Proportional Hazard model
I am doing a study of credit risk in europe for the period of 2006 - 2016 by using the Cox Proportional Hazard Model (time costant edition) in R (coxph). I have succesfully implemented the model for ...
3
votes
0
answers
866
views
Naive Bayes vs. logistic regression [duplicate]
I'm working with credit scoring models.
Here's what I know:
Let Y be the binary outcome variable, $Y \in \{0,1\}$ where $Y = 1$ is the outcome of default and $X = (X_{1},...,X_{m})$ be the random ...
- 619
3
votes
1
answer
1k
views
Optimal classifier or optimal threshold for scoring
In practice, there can be a classifier that gives far better performance at a specific acceptable threshold than an "optimal" classifier with better average performance across range of thresholds (...
- 235
0
votes
0
answers
183
views
Missing Values and Model Scoring
How do you deal with missing values when scoring a model? Can I use multiple imputation when building the model?
- 320
1
vote
1
answer
135
views
How to build Predictive models with insufficient historical/performance data
I'm building a auto loan probability of default model where the loan term could be 3 to 7 years and hence default can happen anytime in that interval. But we are a start-up and have only 3 years of ...
- 79
2
votes
0
answers
65
views
How do loan companies set interest rate tiers?
What statistical or machine learning methods do companies like Lending Club use to segment their customer base into loan grades A1-G5? What would a reasonable partitioning method look like after ...
- 21
1
vote
0
answers
72
views
What is the benefit of developing different scores for LGD modelling?
In the LGD Model flow presented in the figure 4.13 in the book "Developing Credit Risk Models Using SAS Enterprise Miner and SAS/STAT: Theory and Application" which is partially available on the web:
...
- 123
0
votes
1
answer
186
views
Find abnormal credit transactions based on historical data
I have a dataset of customer transactions (multiple customers,multiple transactions) and based on the historical data, I want to know when a new credit(+ve) transaction arrives if its unusual for that ...
- 71