int vs Float in regression modeling This is general question to understand a concept.
I have a dataframe with all columns having float values(precision varies from 2 to 8 digits).
I use GBM to train my model. When i train my model with all float values - r2 score -0.78
Same when all columns are converted to integer values - r2 score -0.72
Why does r2 score drop when float is converted to integer ?
Is it something very specific to my data or in general is it expected to drop ?
 A: Actually, your results make perfect sense.  I would even advance you are lucky.  Yes, you are lucky that your drop in R Square in your model is not much more pronounced.  This phenomenon is I think pretty much related to what is known as the "Butterfly Effect."  This means that a very tiny difference in inputs can make at times a drastic difference in output and model fit.  
When your inputs had a lot of decimals, they were very precise and captured accurately the necessary data for your regression model to capture the relationship between your X variables and your Y variable as accurately as possible.  So, you got a higher R Square. 
When you used integer numbers, you lost quite a bit of information.  You sort of lost some Signal and introduced some Noise in your data.  The rounding creates some incremental random errors in your data set, and your R Square dropped.  That's perfectly normal.  That is what you would expect all the time.  
Inherently, there is nothing wrong with your model and the mentioned divergence in R Square.  If anything, it suggests that your results are reasonably robust because even when using rounded numbers, the R Square holds up pretty well.  
It would be interesting to see if the rounding affected the statistical significance of some of your X variables.  And, if it has... this may warrant additional testing on those X variables.  You could use rolling regressions to see how stable your X variables regression coefficients are. 
