author: Tzu-Ching Wu(George)
This project contains the source code of Kaggle_Customer GUI, a matlab-based GUI for the data preprocessing, feature selection, and classification method testing of the Kaggle competition: Santander Customer Satisfaction. You will obtain the averaged AUC value to evaluate classification performance after running, and can easily create the test dataset result csv file for submitting.
Two matlab-GUI related code: Customer_GUI.m and Customer_GUI.fig are provided. The subfolder /pre_data shoud contains three matlab data file: train.mat, test.mat, and ID.mat which are training dataset, testing dataset, and testing ID from Santander Customer Satisfaction website. Of course, you should have matlab installed in your compyter.
To run Kaggle_Customer-GUI, just type Customer_GUI in matlab command line.
1.Pre-data method:
- Raw data
- Normalization
- Binary
2.Feature selection
- All features
- Choice feature, and input feature numbers
3.Cross-Validation N, input integer number bigger than 2
4.Classification method
- NaiveBayes
- Decision Tree
- Discriminant classification
- K-nearest neighbor(KNN)
- Suppoer vector mechine(SVM)
- Classification Ensemble
- Backpropagation Neural Network(BPN)
- Radial basis network(RBN)
- Adaptive neuro-fuzzy inference system(ANFIS)
v1.1
- Add 4 methods including Classification Ensemble, Backpropagation Neural Network(BPN) Radial basis network(RBN), and Adaptive neuro-fuzzy inference system(ANFIS)
- Show accuracy rate of classification