Statistical Methods for Data Science Mini Project 4

Statistical Methods for Data Science
Mini Project 4
Instructions:
• Total points = 20.
• Submit a typed report.

• Do a good job.
• You must use the following template for your report:
Mini Project #
Name
Names of group members (if applicable)
Contribution of each group member
Section 1. Answers to the specific questions asked
Section 2: R code. Your code must be annotated. No points may be given if a brief
look at the code does not tell us what it is doing.
1. (6 points) In the class, we talked about bootstrap in the context of one-sample problems. But the idea of nonparametric bootstrap is easily generalized to more general
situations. For example, suppose there are two dependent variables X1 and X2 and
we have i.i.d. data on (X1, X2) from n independent subjects. In particular, the data
consist of (Xi1, Xi2), i = 1, . . . , n, where the observations Xi1 and Xi2 come from
the ith subject. Let θ be a parameter of interest — it’s a feature of the distribution
of (X1, X2). We have an estimator ˆθ of θ that we know how to compute from the
data. To obtain a draw from the bootstrap distribution of ˆθ, all we need to do is the
following: randomly select n subject IDs with replacement from the original subject
IDs, extract the observations for the selected IDs (yielding a resample of the original
sample), and compute the estimate from the resampled data. This process can be
1
repeated in the usual manner to get the bootstrap distribution of ˆθ and obtain the
desired inference.
Now, consider the gpa data stored in the gpa.txt file available on eLearning. The
data consist of GPA at the end of freshman year (gpa) and ACT test score (act) for
randomly selected 120 students from a new freshman class. Make a scatterplot of
gpa against act and comment on the strength of linear relationship between the two
variables. Let ρ denote the population correlation between gpa and act. Provide
a point estimate of ρ, bootstrap estimates of bias and standard error of the point
estimate, and 95% confidence interval computed using percentile bootstrap. Interpret
the results. (To review population and sample correlations, look at Sections 3.3.5 and
11.1.4 of the textbook. The sample correlation provides an estimate of the population
correlation and can be computed using cor function in R.)
2. (7 points) Consider the data stored in the file VOLTAGE.DAT on eLearning. These data
come from a Harris Corporation/University of Florida study to determine whether
a manufacturing process performed at a remote location can be established locally.
Test devices (pilots) were set up at both the remote and the local locations and
voltage readings on 30 separate production runs at each location were obtained. In
the dataset, the remote and local locations are indicated as 0 and 1, respectively.
(a) (1 points) Perform an exploratory analysis of the data by examining the distributions of the voltage readings at the two locations. Comment on what you see.
Do the two distributions seem similar? Justify your answer.
(b) (5 points) The manufacturing process can be established locally if there is no
difference in the population means of voltage readings at the two locations. Does
it appear that the manufacturing process can be established locally? Answer this
question by constructing an appropriate confidence interval. Clearly state the
assumptions, if any, you may be making and be sure to verify the assumptions.
(c) (1 point) How does your conclusion in (b) compare with what you expected
from the exploratory analysis in (a)?
3. (7 points) The file VAPOR.DAT on eLearning provide data on theoretical (calculated)
and experimental values of the vapor pressure for dibenzothiophene, a heterocycloaromatic compound similar to those found in coal tar, at given values of temperature.
If the theoretical model for vapor pressure is a good model of reality, the true mean
difference between the experimental and calculated values of vapor pressure will be
zero. Perform an appropriate analysis of these data to see whether or not this is the
case. Be sure to justify all the steps in the analysis.
2