毕业研究笔记3_机器学习公平性实验

Dataset selection:

Dataset1: German Credit Dataset

1. Source: UCI open dataset
2. Link: https://archive.ics.uci.edu/ml/datasets/statlog+(german+credit+data)
3. Sensitive attribute: Gender and Age

Dataset2: Adult Dataset

1. Source: UCI open dataset
2. Link: https://archive.ics.uci.edu/ml/datasets/Adult
3. Sensitive attribute: Gender and Race

Dataset3: Bank marketing Dataset

1. Source: UCI open dataset
2. Link: https://archive.ics.uci.edu/ml/datasets/bank+marketing
3. Sensitive attribute: marital status

Algorithm applied:

1. Logistic regression (LR)
2. Random forest (RF)
3. Neural network (NN)

Experiments (dataset + algorithm model + sensitive attribute)
Experiment set 1
1.1 German + LR + gender
1.2 German + RF + gender
1.3 German + NN + gender
1.4 German + LR + age
1.5 German + RF + age
1.6 German + NN + age

Experiment set 2
2.1 Adult + LR + gender
2.2 Adult + RF + gender
2.3 Adult + NN + gender
2.4 Adult + LR + race
2.5 Adult + RF + race
2.6 Adult + NN + race

Experiment set 3
3.1 Bank + LR + marital
3.2 Bank + RF + marital
3.3 Bank + NN + marital

All experiments will be evaluated with 6 different fairness definitions.