Online NEAT for classification

---

Email: [email protected]

This project uses NEAT (Neuroevolution) to complete the task of 2-classification prediction.  

*Getting Started

This is an example of how you may give instructions on setting up your project locally. To get a local copy up and running follow these simple example steps.

Prerequisites

python3 tensorflow numpy sklearn pandas 

4 types of fitness calculation

Fitness is calculated as follows:

Type	Method
ACC	accuracy
PAN	combines Recall and Specificity fitness=Recall+ Specificity
PRO	utilizes profits of each loan
PAP	fitness(t)=α∙profit(t)+ β∙fitness(t-1)

*Usage

Use this space to show useful examples of how a project can be used. How to run online NEAT

python3 newevolve.py a1 a2 a3 a4  

• a1: The type of Fitness calculation (0-ACC, 1-PAN, 2-PRO, 3-PAP)
• a2: The size of Time window
• a3: α in PAP
• a4: β in PAP

for example: python3 newevolve.py 0 500 0 0

If you want to set other parameters in evolution (like the population size), you can change them in this file: /Neat/config, more explanations can be found in this link

How to run LSTM

python3 LSTM.py 

*About Results

The results will show accuracy, Recall and Specificity in each generation. for example:
0.716 0.9466292134831461 0.14583333333333334
0.722 0.9358288770053476 0.0873015873015873
0.732 0.9523809523809523 0.04918032786885246
....

*About Data

The data is provided by Lendclub

Acknowledgements

• NEAT-python
• Lending Club
• SUSTech HPC which seems to be only available on IE Browser