Build, train and test a machine learning model to predict insurance cost based on customer features such as age, gender, Body Mass Index (BMI), number of children, smoking habits, and geo-location.