I am working on my thesis desertification and I want to apply some econometric model on data. i have a time series data but my dependent variable is count variable in specific number for each year, like number of vehicles registered each month but all the values are under 300. kindly suggest me model . in case you want to help me i can also share my data with anyone.
For count dependent variable you can use poisson model. You may also read "A guide to modern econometrics' by Marno Verbeek for more details.
All the best
Thank you
I think the Poisson model would be a good start then you can consider other models if the Poisson assumptions are violated such Quasi Poisson, Negative Binomial and may be Poisson mixed models.
I don't fully agree with the previous answers. Given that the observations are integers in the range [0 ; 300], I would treat the variable as taking real values. Just consider a transformation (e.g. log or square root) to be sure that forecasts will not be negative.
I agree with Guy Mélard. Just start with a transformation to non-negative and build a time series model (including explanatory variables), treating the dependent variable as continuous.
The Poisson distribution is restrictive, being a one-parameter distribution.
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series