Algorithms in Data Science
Support Vector Machine (SVM)
- Hard margin SVM
- Soft margin SVM
- Non-linear SVM
Decision Tree
- Feature selection
- Tree spanning
- Information gain
- Gini index
- Pruning
Term Frequency - Inverse Document Frequency (TF-IDF)
- Preprocessing (vectorise)
- Term frequency
- $TF(x) = \frac{N(x)}{N}$
- Inverse document frequency
- $IDF(x) = \log \frac{N}{N(x)}$
- Smoothing: $\log \frac{N+1}{N(x)+1}+1$
- $TF-IDF(x) = TF(x) * IDF(x)$
Latent Semantic Analysis (LSA)
- $\arg min_{C_k} X_F = \sqrt{\sum_{i=1}^M \sum_{j=1}^N|X_{ij}|^2},\ where\ X=C-C_k $
- $C=U \Sigma V^T$, $C_k = U\Sigma_kV^T$
Adaptive Boosting (Adaboost) [not clear]
- Steps
- Intilization
- Iteration
- Combine classifications
- Pro and cons
- High accuracy
- Simple, no feature selection
- no overfitting
