Combined PCA and Quantization

Question

I did some PCA on a dataset to reduce the size without compromising too much on their actual information.

However, I learnt that quantization is also an effective technique to do this. I guess it is more fine-grained and leads to compression.

Combining PCA and Quantization will hence be quite effective I guess.

I got PCA to work using Scikit learn. How do I implement Quantization? Are there any libraries or standard methods?

I am working with the Python data science stack.

Thanks for your time. I am not getting how to quantize as it doesn't seem to be a programming issue as I have to change how a number is represented.

Answer 1

I haven't seen usage of quantization in practice, but scikit-learn has several methods that can be used as compression, and trade accuracy for memory/compute time:

• Feature hashing - can be used to extract a specified number of features
• Approximate nearest neighbors with locality sensitive hashing
• Random Trees Embedding - partitions the data using trees and then encodes examples with appropriate indices of leaves