Scribd
Upload
4 views3 pages

Stat Computing 1 Assignment

The document discusses the investigation into the Challenger space shuttle disaster, focusing on the potential link between cold temperatures and O-ring failure. Preliminary analysis indicates a negative relationship between temperature and failure status, supported by logistic regression modeling. The findings suggest that lower temperatures may increase the likelihood of O-ring failures, warranting further exploration through Bayesian models.

Uploaded by

Mrinmoy Banik
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
4 views3 pages

Stat Computing 1 Assignment

The document discusses the investigation into the Challenger space shuttle disaster, focusing on the potential link between cold temperatures and O-ring failure. Preliminary analysis indicates a negative relationship between temperature and failure status, supported by logistic regression modeling. The findings suggest that lower temperatures may increase the likelihood of O-ring failures, warranting further exploration through Bayesian models.

Uploaded by

Mrinmoy Banik
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Statistical Computing Assignment

Mrinmoy Banik
5 December 2024

Introduction
In 1986, the space shuttle Challenger exploded during take off, killing the seven astronauts
on board. The explosion was thought to be the result of an O-ring failure, the splitting of
a ring of rubber that connects parts of the ship together. The accidents was believed to be
caused in the unusually cold weather at the time of launch, as there is reason to believe that
the O-ring failure probability increases and the temperature decreases.
Given some data, we are interested in knowing if the cold temperature was actually
responsible for the crash. We establish some facts from the data:
• Response (Y): Failure status (this is a binary covariate which takes values 1 if a
failure occurs and 0 otherwise)
• Temperature (x): The surrounding temperature (in ◦ F). We scale this covariate to
make our computations easier.

Preliminary Analysis
We first plot the response information against the covariate information to get a clear idea
about the nature of the effect.

1
We can clearly see that there is a negative dependence on temperature: for successful
O-ring experiments, the average scaled temperature is much higher than that for the un-
successful ones. This can be further visualized well by adding some noise to the response
data.

The same conclusion holds.

Frequentist Logistic Regression Model


We fit a frequentist logistic regression model with Y as the response and x as the covariate.
No prior information on the parameters is required. We directly use the glm package in R
to fit a logistic regression model to the data.

2
The summary output shows that the temperature covariate is statistically significant
at the 5% level. The negative sign of the coefficient indicates a negative relationship
between the log-odds of the response (failure status) and the temperature.
To further visualize the relationship, we can plot the failure probability against the scaled
temperature.

 
p
[Note: The logistic regression model is log 1−p
= β0 + β1 x, where p is the probability
exp(β0 +β1 x)
of failure of the O-ring. Rearranging, we get p = 1+exp(β 0 +β1 x)
, which is what is plotted in
the diagram.]
We now try to fit some Bayesian logistic regression models.

You might also like