Disease Prediction
CS 74/174 Project Proposal
Yilong Zhao
April 11, 2012
Problem
Healing illness is the responsibility of a doctor. However, it will be very helpful if patients can
know a little bit about their health before hand. Therefore in this proect, I plan to study and apply several existing
machine learning algorithms on the disease prediction puzzle to achieve the goal that by entering the body information,
people can get a staightforward description of their current health situation. This contributes to the prevention
of a severe disease which may come in silence since people can use it to do daily check.
Data Set Description
I will use the Cleveland database in this project. It has 76 attributes. The training goal refers to the presence of heart
disease in the patient, it is basically an integer from 0 to 4. It is created by Medical Center, Long Beach and Cleveland Clinic Foundation.[1]
Learning Method
Since this project refers to prediction and classification problems. The basic machine learning
algorithms I take into consider are listed as following:
- Decision Tree
- Naive Bayes
- KNN
- SVM
- VSM
- Neural Networks
In order to let the system choose the appropriate method automatically, I plan to implement
some methods to combine the results of those learning algorithms and compare the result of the combination solution.
The ideas are listed below.
- Arbitrarily pick a method - I will pick the method arbitrarily for each test data, I will train all methods on the training data.
- Pick the method with Highest accuracy - I will train all the methods on the training data, and use the method with the
highest accuracy.
- Linear weighted combination - The prediction results of the test data are based on the linear combination of the result of all
the method. The weight of each method-result is based on the accuracy of the method.
Timeline
- April 12: Project proposal;
- April 13 - April 19: Study the related materials, decide which machine learning methods will be used in the project.
- April 20 - May 7: Finish the implementation of the basic methods, get the preliminary result. Milestone object
- May 7 - May 20: Extend the project by implementing more advanced methods.
- May 21 - May 29: Finish all the implementation of the methods, get the combined results and compare them, also finish writing the poster and project write-up.
References