• python 3.6
• sklearn
• pandas
• tensorflow 1.8.0
• keras 2.1.6

• K-Nearest Neighbor (KNN)

• Support Vector Machine (SVM)

• Gaussian Multinomial Naive Bayes (MultinomialNB)

• Decision Tree

• Random Forest

• Extra Tree

• Logistic Regression

• KNN based Bagging

• Logistic Regression

• Majority Voting Ensemble Machine

• Neural network

use first two batches to train && use the last eight batchse to test the results

train accuracy : 0.5200065461091564
cross validation acc   : 0.5200065461091564
batch id: 3  acc: 0.8284993694829761
batch id: 4  acc: 0.6024844720496895
batch id: 5  acc: 0.8629441624365483
batch id: 6  acc: 0.65
batch id: 7  acc: 0.49986161084970937
batch id: 8  acc: 0.20408163265306123
batch id: 9  acc: 0.5127659574468085
batch id: 10  acc: 0.32555555555555554

• 传感器特征的构成
• 数据集的每个传感器的8个维度特征大小是不一样的，所以现在的归一化方式存在问题

1 9 17....
2 10 18 ...
......
8 16 24 ...

分开进行训练