Supervised Learning - Classification Week 7 Challenge

1. PROS AND CONS SVM.

Pros:

1. Effective in High-Dimensional Spaces:

Ø  SVMs perform well in high-dimensional spaces, making them suitable for tasks with a large number of features, such as image classification or text analysis.

2. Robust to Overfitting:

Ø  SVMs have a regularization parameter that helps prevent overfitting, making them robust when dealing with noisy or complex datasets.

3. Versatility:

Ø  SVMs support different kernel functions (linear, polynomial, radial basis function, etc.), allowing them to capture complex relationships and non-linear patterns in the data.

4. Global Optimization:

Ø  SVMs aim to find the hyperplane that maximally separates classes, making them less sensitive to local optima compared to some other algorithms.

5. Effective in Non-Linearly Separable Data:

Ø  SVMs can handle non-linearly separable data by transforming it into a higher-dimensional space through the use of kernel functions.

6. Memory Efficient:

Ø  SVMs use a subset of training points (support vectors) for decision function computation, making them memory-efficient, especially when dealing with large datasets.

7. Mathematically Well-Founded:

Ø  SVMs are based on solid mathematical principles and have well-defined optimization objectives, providing a clear understanding of the model's behavior.

Cons:

1. Sensitivity to Parameters:

Ø  SVMs have parameters, such as the choice of kernel and the regularization parameter (C). The performance of the model can be sensitive to the selection of these parameters.

2. Computational Complexity:

Ø  Training an SVM can be computationally expensive, especially for large datasets. The time complexity is approximately O(n2⋅d), where n is the number of samples and d is the number of features.

3. Interpretability:

Ø  SVMs can be less interpretable compared to simpler models like decision trees. The decision boundaries in high-dimensional spaces may not be easily understandable.

4. Difficulty with Noisy Data:

Ø  SVMs can be sensitive to noisy data, as outliers or mislabeled instances can significantly affect the position of the hyperplane.

5. Memory Intensive for Large Datasets:

Ø  In some cases, SVMs may become memory-intensive, especially when dealing with large datasets or a large number of features.

6. Limited to Binary Classification:

Ø  Traditional SVMs are designed for binary classification tasks. Extensions like one-vs-all can be used for multi-class problems, but they may not perform as well as algorithms specifically designed for multi-class classification.

7. Black Box Model:

Ø  SVMs, especially with non-linear kernels, can be considered as "black box" models, making it challenging to interpret the decision boundaries and feature importance.

2.