Get additional business insight thanks to our AOS solution. Selecting the right KPI will depend on your company and which part of the business you are looking to track. Each department will use different KPI types to measure success based on specific business goals and targets.
HelpAssessment of ARM Success
The fields of all the tables are described below.
Performance and predictions
Name Column | Description |
---|
Model Id | Model identifier |
Date data | Date of the data |
Evaluation date | Date of the evaluation |
Prediction date | Prediction date |
Total customers | Number of customers |
Total products | Number of products |
Total alerts | Number of customers with default prediction |
Total Good | Number of clients with good prediction |
Default ratio | Percentage of clients with predictions of default on the number of clients |
SEN | Sensibility: Percentage of clients with defaults correctly alerted to the total of customers with defaults |
PFA | Probability of false alarms: Percentage of clients alerted erroneously with respect to the total number of clients alerted |
EFF | Efficiency: Percentage of success in predictions |
Product types
Name Column | Description |
---|
Model Id | Model identifier |
Total consumer | Number of consumer loans |
% Consumer | Percentage of consumer products |
M€ consumer | Net value of consumer loans |
Total Mortgages | Number of mortgages |
% Mortgages | Percentage of products that are mortgages |
M€ Mortgages | Net value of mortgages |
Total CC | Number of credit cards |
% CC | Percentage of products that are cards |
M€ CC | Net value of credit cards |
Cost effectiveness
Name Column | Description |
---|
Model Id | Model identifier |
Date data | Date of the data |
Evaluation date | Date of the evaluation |
Prediction date | Prediction date |
M€ alerted | € million in loans alerted |
% € alerted | Percentage of predicted defaults on total net of products |
M€ provisions alerted | Net value of predicted provisions |
% provisions alerted. | Percentage of provisions that have corresponded to defaults |
M€ correctly alerted | Net value of correctly predicted defaults |
M€ false alerts | Net value of the false alarms predicted |
M€ provisions correctly alerted | Net value of provisions predicted with success |
Migrations
Name Column | Description |
---|
Model Id | Model identifier |
Good → Default 1 month | Customers with prediction of "Good" in the last month and "Default" in the current |
Good → Default 2 months | Customers with prediction of "Good" in the penultimate month and "Default" in the current |
Good → Default 3 months | Customers with prediction of "Good" in the third from last month and "Default" in the current |
Default → Good 1 month | Customers with prediction of "Non-payment" in the last month and "Good" in the current |
Default → Good 2 months | Customers with prediction of "Non-payment" in the penultimate month and "Good" in the current |
Default → Good 3 months | Customers with prediction of "Non-payment" in the third from last month and "Good" in the current |
New Customers | Number of clients that were not in the prediction of the previous month but in this one |
Lost Customers | Number of clients that were in the prediction of the previous month but not in this one |