Should I take the entire dataset for Linear Regression Model Training

I am new to the data science and wanted to develop a Linear Regression model to identify important factors to predict a policy holder makes a insurance claim. There are around 750,000 active policy holders in the books and each policy holder details can extract as a row along with factors. What I would like to know, should I take entire 750,000 rows of data and split 70% 30% for training and testing ? Regards, Mira