Selecting train test and validation data for predictive modelling

[anupam3684]

Hi All,

Forgive me for asking such a naive question , but I have got this project on urgent basis and I have no background of stats so for me it is really difficult.

background - I am working on an insurance data (workers compensation) . I have 10 yrs of history data with details like employe name , age  ,  department , injury type , amount_dibursed etc. Now looking into this data I need to predict future trend that how much amount it would be needed if we face same type of injury in future .

Now I have done all the missing value computation .(Thank God) and now need to create test and validation data . I am thinking of 75 : 25 data . Can you please suggest me how to randomly select the data that would best suit me in SAS.

Please do let me know if yuo need more info.

Thanks,

Anupam