| Version: | 1.4-0 |
| Date: | 2016-4-30 |
| Title: | Logistic Joinpoint Regression |
| Author: | Michal Czajkowski, Ryan Gill, Greg Rempala |
| Maintainer: | Ryan Gill <ryan.gill@louisville.edu> |
| Description: | Fits and tests logistic joinpoint models. |
| License: | GPL-2 | GPL-3 [expanded from: GPL (≥ 2)] |
| NeedsCompilation: | yes |
| Packaged: | 2016-05-01 00:09:41 UTC; vbrutac |
| Repository: | CRAN |
| Date/Publication: | 2016-05-01 18:55:59 |
Kentucky yearly cancer mortality from 1999-2005.
Description
This table gives the yearly mortality counts due to neoplasms (ICD 10 codes C00-D48) and population sizes for Kentucky from 1999-2005. For more information, see http://wonder.cdc.gov/wonder/help/cmf.html.
Usage
data(kcm)Format
A 7 by 3 data frame.
Source
Centers for Disease Control and Prevention, National Center for Health Statistics. Compressed Mortality File 1999-2005. CDC WONDER On-line Database, compiled from Compressed Mortality File 1999-2005 Series 20 No. 2K, 2008. Accessed at http://wonder.cdc.gov/cmf-icd10.html on May 5, 2008.
MLE with 0 joinpoints
Description
Determines the maximum likelihood estimate of model coefficients in the logistic joinpoint regression model with no joinpoints.
Usage
ljr0(y,n,tm,X,ofst)
Arguments
y |
the vector of Binomial responses. |
n |
the vector of sizes for the Binomial random variables. |
tm |
the vector of observation times. |
X |
a design matrix containing other covariates. |
ofst |
a vector of known offsets for the logit of the response. |
Details
The re-weighted log-likelihood is the log-likelihood divided by the largest component of n.
Value
Coef |
A table of coefficient estimates. |
wlik |
The maximum value of the re-weighted log-likelihood. |
Author(s)
The authors are Michal Czajkowski, Ryan Gill, and Greg Rempala. The software is maintained by Ryan Gill rsgill01@louisville.edu.
References
Czajkowski, M., Gill, R. and Rempala, G. (2008). Model selection in logistic joinpoint regression with applications to analyzing cohort mortality patterns. Statistics in Medicine 27, 1508-1526.
See Also
Examples
data(kcm)
attach(kcm)
ljr0(Count,Population,Year+.5)
Perform test of 0 vs 1 joinpoints.
Description
This function tests the null hypothesis of 0 joinpoints versus the alternative of one joinpoint based on the likelihood ratio test statistic. The p-value is determined by a Monte Carlo method.
Usage
ljr01(y,n,tm,X,ofst,R=1000,alpha=.05)
Arguments
y |
the vector of Binomial responses. |
n |
the vector of sizes for the Binomial random variables. |
tm |
the vector of ordered observation times. |
X |
a design matrix containing other covariates. |
ofst |
a vector of known offsets for the logit of the response. |
R |
number of Monte Carlo simulations. |
alpha |
significance level of the test. |
Details
The re-weighted log-likelihood is the log-likelihood divided by the largest component of n.
Value
pval |
The estimate of the p-value via simulation. |
Coef |
A table of coefficient estimates. |
Joinpoint |
The estimates of the joinpoint, if it is significant. |
wlik |
The maximum value of the re-weighted log-likelihood. |
Author(s)
The authors are Michal Czajkowski, Ryan Gill, and Greg Rempala. The software is maintained by Ryan Gill rsgill01@louisville.edu.
References
Czajkowski, M., Gill, R. and Rempala, G. (2008). Model selection in logistic joinpoint regression with applications to analyzing cohort mortality patterns. Statistics in Medicine 27, 1508-1526.
See Also
Examples
data(kcm)
attach(kcm)
set.seed(12345)
## Not run: ljr01(Count,Population,Year+.5,R=20)
MLE with 1 joinpoint
Description
Determines the maximum likelihood estimates of model coefficients in the logistic joinpoint regression model with one joinpoint.
Usage
ljr1(y,n,tm,X,ofst,summ=TRUE)
Arguments
y |
the vector of Binomial responses. |
n |
the vector of sizes for the Binomial random variables. |
tm |
the vector of ordered observation times. |
X |
a design matrix containing other covariates. |
ofst |
a vector of known offsets for the logit of the response. |
summ |
a boolean indicator of whether summary tables should be returned. |
Details
The re-weighted log-likelihood is the log-likelihood divided by the largest component of n.
Value
Coef |
A table of coefficient estimates. |
Joinpoint |
The estimate of the joinpoint. |
wlik |
The maximum value of the re-weighted log-likelihood. |
Author(s)
The authors are Michal Czajkowski, Ryan Gill, and Greg Rempala. The software is maintained by Ryan Gill rsgill01@louisville.edu.
References
Czajkowski, M., Gill, R. and Rempala, G. (2008). Model selection in logistic joinpoint regression with applications to analyzing cohort mortality patterns. Statistics in Medicine 27, 1508-1526.
See Also
Examples
data(kcm)
attach(kcm)
ljr1(Count,Population,Year+.5)
Test coefficients conditioned on K=1 joinpoint.
Description
This function performs the likelihood ratio tests to find p-values in testing the significance of each of the coefficients as well as the intercept and ordered observation times. The p-values are determined by a Monte Carlo method.
Usage
ljr11(y,n,tm,X,ofst,R=1000)
Arguments
y |
the vector of Binomial responses. |
n |
the vector of sizes for the Binomial random variables. |
tm |
the vector of ordered observation times. |
X |
a design matrix containing other covariates. |
ofst |
a vector of known offsets for the logit of the response. |
R |
number of Monte Carlo simulations. |
Details
The re-weighted log-likelihood is the log-likelihood divided by the largest component of n.
Value
pvals |
The estimates of the p-values via simulation. |
Author(s)
The authors are Michal Czajkowski, Ryan Gill, and Greg Rempala. The software is maintained by Ryan Gill rsgill01@louisville.edu.
References
Czajkowski, M., Gill, R. and Rempala, G. (2008). Model selection in logistic joinpoint regression with applications to analyzing cohort mortality patterns. Statistics in Medicine 27, 1508-1526.
See Also
Examples
data(kcm)
attach(kcm)
set.seed(12345)
## Not run: ljr11(Count,Population,Year+.5,R=20)
Perform backward joinpoint selection algorithm with upper bound K.
Description
This function performs the backward joinpoint selection algorithm with K maximum possible number of joinpoints based on the likelihood ratio test statistic. The p-value is determined by a Monte Carlo method.
Usage
ljrb(K,y,n,tm,X,ofst,R=1000,alpha=.05)
Arguments
K |
the pre-specified maximum possible number of joinpoints |
y |
the vector of Binomial responses. |
n |
the vector of sizes for the Binomial random variables. |
tm |
the vector of ordered observation times. |
X |
a design matrix containing other covariates. |
ofst |
a vector of known offsets for the logit of the response. |
R |
number of Monte Carlo simulations. |
alpha |
significance level of the test. |
Details
The re-weighted log-likelihood is the log-likelihood divided by the largest component of n.
Value
pvals |
The estimates of the p-values via simulation. |
Coef |
A table of coefficient estimates. |
Joinpoints |
The estimates of the joinpoint, if it is significant. |
wlik |
The maximum value of the re-weighted log-likelihood. |
Author(s)
The authors are Michal Czajkowski, Ryan Gill, and Greg Rempala. The software is maintained by Ryan Gill rsgill01@louisville.edu.
References
Czajkowski, M., Gill, R. and Rempala, G. (2008). Model selection in logistic joinpoint regression with applications to analyzing cohort mortality patterns. Statistics in Medicine 27, 1508-1526.
See Also
Examples
data(kcm)
attach(kcm)
set.seed(12345)
## Not run: ljrb(1,Count,Population,Year+.5,R=20)
Perform forward joinpoint selection algorithm with unlimited upper bound.
Description
This function performs the full forward joinpoint selection algorithm based on the likelihood ratio test statistic. The p-value is determined by a Monte Carlo method.
Usage
ljrf(y,n,tm,X,ofst,R=1000,alpha=.05)
Arguments
y |
the vector of Binomial responses. |
n |
the vector of sizes for the Binomial random variables. |
tm |
the vector of ordered observation times. |
X |
a design matrix containing other covariates. |
ofst |
a vector of known offsets for the logit of the response. |
R |
number of Monte Carlo simulations. |
alpha |
significance level of the test. |
Details
The re-weighted log-likelihood is the log-likelihood divided by the largest component of n.
Value
pvals |
The estimates of the p-values via simulation. |
Coef |
A table of coefficient estimates. |
Joinpoints |
The estimates of the joinpoint, if it is significant. |
wlik |
The maximum value of the re-weighted log-likelihood. |
Author(s)
The authors are Michal Czajkowski, Ryan Gill, and Greg Rempala. The software is maintained by Ryan Gill rsgill01@louisville.edu.
References
Czajkowski, M., Gill, R. and Rempala, G. (2008). Model selection in logistic joinpoint regression with applications to analyzing cohort mortality patterns. Statistics in Medicine 27, 1508-1526.
See Also
Examples
data(kcm)
attach(kcm)
set.seed(12345)
## Not run: ljrf(Count,Population,Year+.5,R=20)
Perform test of j vs k joinpoints.
Description
This function tests the null hypothesis of j joinpoint(s) versus the alternative of k joinpoint(s) based on the likelihood ratio test statistic. The p-value is determined by a Monte Carlo method.
Usage
ljrjk(j,k,y,n,tm,X,ofst,R=1000,alpha=.05)
Arguments
j, k |
pre-specified number of joinpoints in the null and alternative hpyotheses (the smaller is used for the null). |
y |
the vector of Binomial responses. |
n |
the vector of sizes for the Binomial random variables. |
tm |
the vector of ordered observation times. |
X |
a design matrix containing other covariates. |
ofst |
a vector of known offsets for the logit of the response. |
R |
number of Monte Carlo simulations. |
alpha |
significance level of the test. |
Details
The re-weighted log-likelihood is the log-likelihood divided by the largest component of n.
Value
pval |
The estimate of the p-value via simulation. |
Coef |
A table of coefficient estimates. |
Joinpoint |
The estimates of the joinpoint, if it is significant. |
wlik |
The maximum value of the re-weighted log-likelihood. |
Author(s)
The authors are Michal Czajkowski, Ryan Gill, and Greg Rempala. The software is maintained by Ryan Gill rsgill01@louisville.edu.
References
Czajkowski, M., Gill, R. and Rempala, G. (2008). Model selection in logistic joinpoint regression with applications to analyzing cohort mortality patterns. Statistics in Medicine 27, 1508-1526.
See Also
Examples
data(kcm)
attach(kcm)
set.seed(12345)
## Not run: ljrjk(0,1,Count,Population,Year+.5,R=20)
MLE with k joinpoints
Description
Determines the maximum likelihood estimates of model coefficients in the logistic joinpoint regression model with k joinpoints.
Usage
ljrk(k,y,n,tm,X,ofst)
Arguments
k |
the pre-specified number of joinpoints (with unknown locations). |
y |
the vector of Binomial responses. |
n |
the vector of sizes for the Binomial random variables. |
tm |
the vector of ordered observation times. |
X |
a design matrix containing other covariates. |
ofst |
a vector of known offsets for the logit of the response. |
Details
The re-weighted log-likelihood is the log-likelihood divided by the largest component of n.
Value
Coef |
A table of coefficient estimates. |
Joinpoints |
The estimates of the joinpoints. |
wlik |
The maximum value of the re-weighted log-likelihood. |
Author(s)
The authors are Michal Czajkowski, Ryan Gill, and Greg Rempala. The software is maintained by Ryan Gill rsgill01@louisville.edu.
References
Czajkowski, M., Gill, R. and Rempala, G. (2008). Model selection in logistic joinpoint regression with applications to analyzing cohort mortality patterns. Statistics in Medicine 27, 1508-1526.
See Also
Examples
data(kcm)
attach(kcm)
ljrk(1,Count,Population,Year+.5)
Test coefficients conditioned on K=k joinpoint.
Description
This function performs the likelihood ratio tests to find p-values in testing the significance of each of the coefficients as well as the intercept and ordered observation times. The p-values are determined by a Monte Carlo method.
Usage
ljrkk(k,y,n,tm,X,ofst,R=1000)
Arguments
k |
the pre-specified number of joinpoints (with unknown locations). |
y |
the vector of Binomial responses. |
n |
the vector of sizes for the Binomial random variables. |
tm |
the vector of ordered observation times. |
X |
a design matrix containing other covariates. |
ofst |
a vector of known offsets for the logit of the response. |
R |
number of Monte Carlo simulations. |
Details
The re-weighted log-likelihood is the log-likelihood divided by the largest component of n.
Value
pvals |
The estimates of the p-values via simulation. |
Author(s)
The authors are Michal Czajkowski, Ryan Gill, and Greg Rempala. The software is maintained by Ryan Gill rsgill01@louisville.edu.
References
Czajkowski, M., Gill, R. and Rempala, G. (2008). Model selection in logistic joinpoint regression with applications to analyzing cohort mortality patterns. Statistics in Medicine 27, 1508-1526.
See Also
Examples
data(kcm)
attach(kcm)
set.seed(12345)
## Not run: ljrkk(1,Count,Population,Year+.5,R=20)