homework 3 explain

part1:

idea of the my.grad.desc()

my.grad.desc() X_n+1  <- X_n - lamda*f`(x) #(lamda bi small)

# number of iter = large
nln() = non.linear minimization
newton's X_n+1 <- X_n - (f``(x_n))^-1 * f(`(x_n))
lamda = 1/f(``(Xn))

# number of iter = small 
# cost per iter is high

knn:

• turning parameter traning set is a par___ that er nomarly "true" to minimide test effects be like:

• k-fold CV error estimated with several times, like:

  

• after that ,we say we fitting the k-fold error set

Tidyverse

• several packages like:
  • dpylr
  • ggplot2
  • reader
  • tibble: of the data.frame
  • tidyr
  • purrr
• and we weill focus on: (see slides and recording)

Importing Data with Base R, readr, and fread

• readr provides faster tabular inporting framework
• fread is not asscesed: not a part of the tidyverse