首页 > > 详细

data frames辅导、讲解R编程语言、辅导R设计、UCI Machine讲解调试Matlab程序|讲解R语言编程

Assigned:
November 17, 2019 Homework 4
Due:
December 01, 2019
Please complete the assigned problems to the best of your abilities. Ensure
that the work you do is entirely your own, external resources are only used as
permitted by the instructor, and all allowed sources are given proper credit for
non-original content.
1 Recitation Exercises
These exercises are to be found in: Introduction to Statistical Learning,
1.1 Chapter 8
Exercises: 1,3,4,5
1.2 Chapter 10
Exercises: 1,2,3,4,6
2 Practicum Problems
These problems will primarily reference the lecture materials and the examples
given in class using R and CRAN. It is suggested that a RStudio session be
used for the programmatic components.
2.1 Problem 1
Simulate a binary classification dataset with a single feature via a mixture of
normal distributions using R (Hint: Generate two data frames with the random
number and a class label, and combine them together). The normal distribution
parameters (using the function rnorm) should be (5,2) and (-5,2) for the pair of
samples - you can determine an appropriate number of samples. Induce a binary
decision tree (using rpart), and obtain the threshold value for the feature in
the first split. How does this value compare to the empirical distribution of
the feature? How many nodes does this tree have? What is the entropy and
Gini at each? Repeat with normal distributions of (1,2) and (-1,2). How many
nodes does this tree have? Why? Prune this tree (using rpart.prune) with a
complexity parameter of 0.1. Describe the resulting tree.
2.2 Problem 2
Load the Wine sample dataset from the UCI Machine Learning Repository
(wine.data) into R using a dataframe (Note: The column names will need
to be loaded separately). Use either the prcomp or princomp methods to
perform a PCA of the wine data - justify whether scaling of the inputs should
Prof. Panchal:
Wed. 6:25PM-9:05PM
MATH 571 - Data Preparation & Analysis Fall 2019:
Section 01-03
Assigned:
November 17, 2019 Homework 4Due:
December 01, 2019
be used or not when performing the decomposition. Plot a biplot of the results
- identify a feature which is pointed in the opposite direction of Hue in
principal component/rotated feature space. What does this imply regarding the
correlation of this feature to Hue? Support your result with a calcualted value.
Finally, plot a screeplot of your results and determine the percentage of total
variance explained by PC1 and PC2.
2.3 Problem 3
Load the USArrests sample dataset from the built-in datasets (data(USArrests))
into R using a dataframe (Note: Row names are states, not numerical values!).
Use the kmeans package to perform a clustering of the data with increasing
values of k from 2 to 10 - you will need to decide whether or not to center/scale
the observations - justify your choice. Plot the within-cluster sum of squares for
each value of k - what is the optimal number of clusters? Use the tidyverse and
fviz cluster plotting method from factoextra to plot the optimal clustering.
2.4 Problem 4
Load the Wine Quality sample dataset from the UCI Machine Learning Repository
(winequality-white.csv) into R using a dataframe (Note: There is both
a red and white wine file, we will use white!). Excluding the quality target
variable, use hclust to perform a hierarchical clustering of the data with single
as well as complete linkage. You will need to decide on whether or not
to center/scale the observations - justify your choice. At what distance value
are the two penultimate clusters merged? Use the cutree method to obtain
these two clusters, and calculate their summary statistics. What feature means
have the largest differences? Which linkage method produces are more balanced
clustering?
Prof. Panchal:
Wed. 6:25PM-9:05PM
MATH 571 - Data Preparation & Analysis Fall 2019:Section 01-03

联系我们 - QQ: 99515681 微信:codinghelp
程序辅导网!