In this project,I explored this dataset. This is basically a classification problem. I have tried various classical machine learning models like SVM, K-NN, Desicion tree and logistic regression to get state-of-the-art accuracy.Finally I have also used deep learning models like multilayer perceptron and 1D conv-net. Uniqueness of the datset set is that it has 512 features that is the feature space has 512 dimensions.
Result of the predictive models:
Model | Accuracy |
---|---|
logistic regression | Test Accuracy: 0.9500316255534472 |
MLP | Test accuracy: 0.932 |
1D Conv | Test accuracy: 0.917 |
SVM ker = rbf | Test accuracy: 0.941808981657179 |
SVM ker = linear | Test accuracy: 0.9487666034155597 |
SVM ker = Polynomial | Test accuracy: 0.92662871600253 |
KNN | Test set Accuracy: 0.8548387096774194 |
Decision Tree | DecisionTrees's Test Accuracy: 0.8358633776091081 |