A medical diagnosis system built from 10,000 broken (1 attribute missing) patient records using Bayesian Networks.
- problem.pdf - The problem statement
- alarm.bif - Describes the whole bayesian network of diseases and symptoms.
- records.dat - Contains all the patient records.
- gold_alarm.bif - Contains ideal probability values.
To run -
bash compile.sh
bash run.sh alarm.bif <sample_data>.dat
It will generate a file solved_alarm.bif.
To check the correctness -
g++ -o format_check Format_checker.cpp
./format_check
It will output a score which is a measure of error from the gold values.