0
$\begingroup$

What other techniques can be used in place of a Logistic Regression model?

Also is there any other method besides MLE for estimating the Logistic Regression parameters?

$\endgroup$
7
  • 3
    $\begingroup$ What is the objective of your modeling? $\endgroup$
    – whuber
    Dec 4, 2012 at 18:20
  • $\begingroup$ What data have you got? In particular, what's the nature of your dependent variable (if you have one). $\endgroup$
    – Peter Flom
    Dec 4, 2012 at 18:22
  • $\begingroup$ I have binary data i.e. 0,1 dependant variable and the explanatory variables are both categorical and continuous $\endgroup$
    – Mukul
    Dec 4, 2012 at 18:28
  • $\begingroup$ I want to predict whether a customer will churn or not and what factors the churn might depend on $\endgroup$
    – Mukul
    Dec 4, 2012 at 18:53
  • 2
    $\begingroup$ Any reason you feel logistic regression isn't sufficient for the job? $\endgroup$
    – AdamO
    Dec 4, 2012 at 20:11

3 Answers 3

1
$\begingroup$

You could use Bayesian models, instead of pure MLE, the MCMCpack or the Zelig library in R will fit Bayesian logistic regression models

$\endgroup$
1
$\begingroup$

This is a classification problem. Machine learning has a lot of tools to address problems such as these, e.g., Neural Networks, Support Vector Machines (SVM), Classification and Regression Trees (CART) etc.

$\endgroup$
1
$\begingroup$

Consider survival analysis, such as cox regression, because time to churn is also important. You can also consider exponential, gamma, weibull models along the lines of survival analysis.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.