On The Joys Of Re-reading Our Old Books
Since I first wrote on this subject in my Facebook, (see image of text of the FB post above), I have discovered yet another classic: Neural, Novel and Hybrid Algorithms for Time Series prediction~ by Timothy Masters [Wiley, 1995]. It's eighteen years since the book was first published. But the topics covered (especially on data pre-processing, scaling and transformations) are still as important. I read each page with the new advantage of perspectives and practical experience gained on this subject these past eighteen years. Old words took on a new significance, a deeper meaning, and sometimes a different meaning. I begin to see the advantages of of hybrid models that combine Neural Nets, Genetic Algorithms, Statistics and Digital Signal Processing. Here are the Contents of this book: 1. Preprocessing 2. Subduing Seasonal Components 3. Frequency-Domain Techniques I 4. Frequency-Domain Techniques II 5. Wavelet and QMF Features 6. Box-Jenkin ARMA Models 6. Differ...