Beaver is an open-source application to visualize and validate the behavior of dynamic models of credit risk. The target users of our software are researchers, financial institutions, and other end users interested in credit analysis. Credit analysis is an important research direction in the finance field. Up to now, there is almost no free professional software and instructions available. Our Beaver software is not only for validate and visualize the effectiveness of dynamic survival models. Above all, it is an effective tool to take the economic environment and different datasets into consideration. Moreover, each end-user could use their account to ensure data privacy. Besides, to help users' future analysis, visualization graphs generated under different parameters could be overlapped together for comparison. We will discuss more other features in later parts. All these work aim to make the tool robust and user-friendly, without assuming the end-user will be very computer literate.

Beaver contains the following features:

• Users can register/ recover password/ log in/ log out when they want to use/ exit the software

Sign in	Register account	Recover password

• Users can import their data after setting the filter condition and select the appropriate data set to apply to a certain model application operation
• Users can choose a model they want from a model list and add new models they have built in Python for experiment
• Users can do the stress testing and select different test scenarios to observe the effect of the stress test on the target data set
• Users can get the model application result in the form of data visualization (e.g. Default Rate/ Log-Likelihood/ Scenario Chart) and get a statistic dashboard showing the economic environment

Default Rate Graph	Log - Likelihood Graph

• Users can save/ export the test results obtained from the current model application operation and merge graphs under different parameters to get a complete comparison graph

Default Rate Merge Graph	Log - Likelihood Merge Graph

• Users can browse/ rename/ delete the model/ dataset/ scenario and mark which are favored

1. Contributors
2. Arrangement for Meetings
3. Operating System
4. Setup Environment Variables
5. Install Beaver
6. Class Diagram
7. Class Description
8. Design Pattern
9. License

Beaver is designed based on Windows X64 operating system and has not been deployed on Linux or Macintosh OS systems.

Beaver has been deployed and run on Windows 10 X64.

To successfully run Beaver, following environment variables need to be installed:

• Java Runtime Environment version 1.8

Beaver is developed as a Java project and need to be run on Java Runtime Environment 1.8 (JRE 1.8) or later versions. Or user can can install the Java Development Kit 11 (JDK 11) or later version as JRE 1.8 is included in it. User can download the above from Java Software | Oracle and choose required products.

• Python

Because dynamic models are built in Python, which means the core part of credit risk analysis is implemented in Python. Java Runtime method is used to execute Python script, which makes code productive and dynamic. The main idea of Java Runtime method is to execute Python script through command line and get the execution result from the script output stream that should have been printed in the console. However, some compulsory third-party libraries are not available. As a result, you need to install the required libraries manually. To do this, carry out the following steps:

pip install numpy
pip install sklearn
pip install pymysql

Attention: If you use other third party libraries in Python model script, install these libraries in the same way.

1. Download Beaver form http://cslinux.nottingham.edu.cn/~Team202001/
2. Follow the guide step by step to set the installation settings
3. Finish installation

We have used several design patterns (e.g. Model - View - Controller (MVC), Singleton) to make our code more organized and easy for us to maintain. These kinds of design patterns could help developers write code faster by providing a picture of how you are implementing the design. Besides, it encourages code reuse and accommodates change by supplying well-tested mechanisms for delegation and composition.

Singleton restricts the number of instantiation of a class, which means only one instance of the class exists in the java virtual machine. A Singleton encapsulates a unique resource and makes it readily available throughout the application.

Singleton	Singleton Class

MVC is an architecture or a software design pattern that makes creating huge applications easy. It separates user interface and related program logic into three interconnected elements: Model, View, and Controller. Model is the central part of the program, it is responsible for data processing and calculation logic. View is where the user operates directly, it may reflect the calculation results of model calculation through the form of table and chart. Controller is the bridge between model and view, it accepts input and converts it to commands for the model or view.

MVC	Model	View	Controller

MIT