Problem 1. Implement logistic corruption with both the "accuracy" and the "binary cantankerous entropy" cold functions, and abstraction how the after-effects and the achievement of the algorithm differ.
Problem 2. Implement backbone corruption with assorted ethics of the regularization parameter, and additionally analyze with application pseudo-inverse with assorted ethics of the blow parameter. Try them both on the abstracts set area the ascribe capricious is a agent of ambit two, area the both entries (and the achievement variables) are according to i + N(0.1, 1), area i goes from 0 to 100000.
Problem 3. (a added astute adaptation of botheration 2). Use the inputs the day to day changes of the prices of the four banal indices (DJIA, S&P 500, Russell 2000, and Nasdaq 100) and as achievement the change in amount of MSFT. Have your assay run over the aeon of the aftermost 20 years. Figuring out area to get the abstracts is allotment of the project.
Problem 4. Do variants of botheration 3 for altered one-year-long periods in history to see if the statistics change over time.
Order a unique copy of this paper