• Gauhati University

This paper shows a mathematical modeling method using different machine learning algorithms for prediction of
probability of procuring Pancreatic Cancer (PC). Each algorithm reports its own accuracy, precision, recall and
F1-score. Also, a Bayesian network model is used to determine the probability each subject has in contracting PC
on the basis of certain preconditions, like his dietary habits and other biological attributes. This paper makes use of
the PC dataset as provided by the National Cancer Institute in collaboration with National Institute of Health (NIH).
The features obtained from this dataset can have either a binary value or a scalar value. The dataset consists of three questionnaires distributed to 155000 subjects. In each of these questionnaires, the subject is asked about his dietary habits and illness history.