数据集概述

数据集名称

• Title / Machine Learning Model Test
• Credit Approval Dataset

数据集来源

• UCI Machine Learning Lab
• UCI data repository

数据集链接

• Credit Approval Dataset

数据集内容

• 属性数量：16个（A1至A16）
• 数据类型：包含连续型和非连续型数据
• 缺失值：存在以“?”表示的缺失值
• 目标数据：二元分类，以“+”和“-”形式出现在A16中

数据预处理步骤

1. 数据清洗：处理以“?”表示的缺失值
2. 数据类型转换：将A2, A3, A8, A11, A14, A15的连续型数据转换为float类型
3. 目标数据转换：将A16中的“+”和“-”转换为0和1

模型评估方法

• 10-fold交叉验证
• 评估指标：Accuracy, Precision, Recall, F-1 Score

模型评估结果

• 最高准确度：Random Forest模型，达到0.902
• 最高精确度：Random Forest模型，达到0.855
• 最高召回率：CNN模型，达到0.870
• 最高F1分数：CNN模型，达到0.830

结论

• 最优模型：Random Forest, CNN, Decision Tree
• 性能较低模型：Multilayer Perceptron

参考文献

1. Bhukya, D. and Ramachandram, S. (2010). Decision Tree Induction: An Approach for Data Classification Using AVL-Tree. International Journal of Computer and Electrical Engineering, pp.660-665.
2. Chen, L. and Tang, H. (2004). Improved computation of beliefs based on confusion matrix for combining multiple classifiers. Electronics Letters, 40(4), p.238.
3. Fourie, C. (2003). Deep learning? What deep learning?. South African Journal of Higher Education, 17(1).
4. Koo, I., Lee, N. and Kil, R. (2008). Parameterized cross-validation for nonlinear regression models. Neurocomputing, 71(16-18), pp.3089-3095.
5. Mantas, C., Castellano, J., Moral-García, S. and Abellán, J. (2018). A comparison of random forest based algorithms: random credal random forest versus oblique random forest. Soft Computing, 23(21), pp.10739-10754.
6. Mühlenbein, H. (1990). Limitations of multi-layer perceptron networks - steps towards genetic neural networks. Parallel Computing, 14(3), pp.249-260.
7. Uchida, K., Tanaka, M. and Okutomi, M. (2018). Coupled convolution layer for convolutional neural network. Neural Networks, 105, pp.197-205.