Influential Parameters

Hi All,

I have small doubt regarding the weights coming from the gradient boosting algorithm.i have a first model with small data set and got the weights from gradient boosting for one feature lets say attribute 1 ( which i am interested) the weight is low and again after some days i got a new data set where i have some additinal data and i when ran the model i got high weight for attribute 1 .Now i want to know what change in the new data made that change in weight and it became high influencing parameter. Do i need to look at data or is there any way i can trouble shoot this to figure out ?

Regards,

Vishnu