In: Statistics and Probability
For what type of dependent variable are poisson and negative binomial regressions appropriate? Give an example of such a variable. In what metric are these regression coefficients? What can we do to them to make them more interpretable, and how would we interpret the resulting translated coefficients?
(Understanding and Using Statistics for Criminology and Criminal Justice)
Answer:
Given Data
In case of logistic regression the dependent variable must be a dichotomous variable
i.e it will take only two values.
Let be the dependent variable , it can be defined as
For logistic regression will be a categorial variable and there will be only 2 categories
Metric for logistic regression coefficiennt.
True values of varible.
Predicted values of variable based on the explanatory variables.
Construct a confusion Matrix
predicted | |||
0 | 1 | ||
True | 0 | a | b |
1 | c | d |
From the above table we can see there are total n observation.
Out of these (a+d) many are classified correctly and (b+c) many are classified wrongly.
Accuracy is defined as a measure of the logistic regression.
Accuracy =
The higher the accuracy the better the classification
Interpretation of coefficient:
Let be the estimated coefficient in the logistic regression.
Define odds ratio =
the odds ratio can be interpreted as the estimated increase in the probability of sucess associated with one unit change in the value of the explanatory variables.
**Please like it..