Scribd
Upload
4 views1 page

Logistic and Cox Regression Explained

The document discusses logistic regression, which analyzes the relationship between the log odds of a binary event and various predictor variables, providing odds ratios for interpretation. It also covers Cox regression, a multivariate technique that estimates hazard ratios based on time-to-event data while accounting for censoring. Both methods include formulas and emphasize the importance of interaction terms and model fit measures like the area under the ROC curve.

Uploaded by

Sara Aspuru
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 views1 page

Logistic and Cox Regression Explained

The document discusses logistic regression, which analyzes the relationship between the log odds of a binary event and various predictor variables, providing odds ratios for interpretation. It also covers Cox regression, a multivariate technique that estimates hazard ratios based on time-to-event data while accounting for censoring. Both methods include formulas and emphasize the importance of interaction terms and model fit measures like the area under the ROC curve.

Uploaded by

Sara Aspuru
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

Unit

 9  
• Logistic  regression  
o Logistic  regression  calculates  the  linear  dependence  between  the  log  odds  of  a  
binary  event  and  one  or  more  continuous  or  binary  variables  
o Formula:  

 
o Exponentiating  the  betas  in  the  equation  gives  odds  ratios.  The  odds  ratios  give  
the  increase  in  odds  for  a  1-­‐unit  increase  in  the  predictor,  adjusted  for  all  other  
variables  in  the  model.  
o The   area   under   the   ROC   curve   is   a   measure   of   the   fit   of   the   model   (50%   -­‐   no  
predictive  ability)  
o Interaction:  When  the  effect  size  is  significantly  different  in  different  sub-­‐groups.  
To  test  for  interactions,  add  an  interaction  term  to  the  model.  A  significant  beta  
for  the  interaction  term  indicates  significant  interaction  
• Cox  regression  
o Cox  regression  is  the  multivariate  regression  technique  where  the  time-­‐to-­‐event  
(taking  censoring  into  account)  is  the  dependent  variable  
o Cox  regression  estimates  hazard  ratios,  a  ratio  of  hazard  rates  (a  hazard  rate  is  
the  probability  of  an  event  occurring  at  the  next  instant  given  it  hasn’t  occurred  
already)  
o Formula:  
 
o The   exponentials   of   βj   in   the   equation   quantifies   the   hazard   ratio   of   increasing  
the  xj  by  1  unit,  controlling  for  the  other  variables  in  the  model  
 

 
 
 

You might also like