Deloitte Data Analytics

Deloitte Data Analytics

1. Ingredients

   • Distributed files systems
   • Database management
   • Advanced Analytics
   • Case study

2. Credit Scoring model

   • Health Score
     • Probability of missing a payment
   • New Clients
     • Respect to financial obligations
   • What drives default

3. Data Quality

   • Preprocessing and checking the quality of the underlying data
   • Strong Modelling Challenges with data of 5% defaults
   • Can be numerical or categorical
   • Outliners events that occur with a small probability and a high impact, they are a result of system error.
     • Can be detected by the use of boxplots
   • There are ways to impute missing values, such as the expectation-maximisation algorithm.
   • Quality assurance: Inconsistency, Incompleteness, Accuracy, Precision, Missing / Unknown.

4. Model Development

   • Definition is important and the core of the prediction of the model
   • Classification is key as true population and the noise.
   • Reject inference
   • Logistic Regresion
     • This function that takes as input the client characteristics and outputs the probability of default
   • Generalisation error
     • Extracting the correct coefficients by minimising the error between model output and observed output
     • Test the ability to give the correct answer to a new case

5. Model performance

   • Goodness of fit
   • Predictive power
   • Confussion Matrix

6. Model Refirements

7. Model interpretation