What are the differences between PCA and t-SNE?

What are the differences between PCA and t-SNE?

PCA (Principal Component Analysis) is a linear dimensionality reduction technique, while t-SNE (t-distributed Stochastic Neighbor Embedding) is a non-linear technique. PCA is often used for reducing dimensions while preserving variance, whereas t-SNE is great for visualizing high-dimensional data by preserving local structures.

When should I use PCA instead of t-SNE?

Use PCA when you need to reduce dimensions for further analysis or when your data is linearly separable. t-SNE is better for visualizing complex, high-dimensional data where you want to see clusters.

Can you give examples of use cases for both?

Sure! PCA is commonly used in image compression and gene expression analysis. t-SNE is often used in visualizing clusters in datasets like MNIST or in single-cell RNA sequencing.

Are there any limitations to using t-SNE?

Yes, t-SNE can be computationally intensive and may not preserve global structures well. It's also sensitive to parameter settings.