This repository contains the python implementation for the expirment part in the master thesis, and also the datasets the has been used, for replication purposes.

• Abstract
• Screenshots
• Technologies
• Setup
• Features
• Status
• Inspiration
• Contact

Survival analysis is a hotspot in statistical research for modelling time-to-event information with data censorship handling, which has been widely used in many applications such as clinical research, information system and other fields with survivorship bias. Many works have been proposed for survival analysis ranging from traditional statistic methods to machine learning and deep learning methods. This paper examines novel deep learning techniques for survival analysis in credit risk modelling context. After surveying through literature for deep learning survival analysis models in various domains and categorizing them, we evaluate the adequacy of six different models representing different categories, using two datasets of US mortgages from separate sources. The performance of these models is evaluated using the discrimination metric, concordance index.

The following package versions have been used to develop this work.

python 3.7.9
lifelines==0.25.4
pandas==1.1.4
pycox==0.2.1
scikit-learn==0.24.1
torch==1.7.0
matplotlib==3.3.3

DATE and Deephit:
tensorflow==1.15.0

DRSA: 
tensorflow==2.0.0

We consider the following datasets:

• M1 (U.S mortgage data provided by International Financial Research)
• M2 (single-family US mortgage loans colected by Blumenstock et al., 2020 [1]).

The datasets directory contains M1 dataset (named as mortgage) and ten batches from M2 (named as data batches).

• Tech 1 - version 1.0
• Tech 2 - version 2.0
• Tech 3 - version 3.0

• Tech 1 - version 1.0
• Tech 2 - version 2.0
• Tech 3 - version 3.0

Describe how to install / setup your local environement / add link to demo version.

Show examples of usage: put-your-code-here

List of features ready and TODOs for future development

• Awesome feature 1
• Awesome feature 2
• Awesome feature 3

To-do list:

• Wow improvement to be done 1
• Wow improvement to be done 2

Project is: in progress, finished, no longer continue and why?

Add here credits. Project inspired by..., based on...

Created by @flynerdpl - feel free to contact me!

[1] Gabriel Blumenstock, Stefan Lessmann & Hsin-Vonn Seow (2020). Deep learning for survival and competing risk modelling. Journal of the Operational Research Society. [paper]

[2] Håvard Kvamme, Ørnulf Borgan, and Ida Scheel. Time-to-event prediction with neural networks and Cox regression. Journal of Machine Learning Research, 20(129):1–30, 2019. [paper]

[3] Jared L. Katzman, Uri Shaham, Alexander Cloninger, Jonathan Bates, Tingting Jiang, and Yuval Kluger. Deepsurv: personalized treatment recommender system using a Cox proportional hazards deep neural network. BMC Medical Research Methodology, 18(1), 2018. [paper]