Comparative Study of Feature Engineering Techniques for Disease Prediction

Feature engineering is essential for desigining predictive models using online text. To fit appropriate machine learning models for text analysis, feature extraction and selection is need to be done rightfuly. This paper presents a comparative study of a number of feature extraction and feature selection techniques useful for text analysis and also presents a feature selection technique inspired from the existing methods. In particular the problem focused here is predicting diseases based on symptoms descriptions collected from online free text. A good number of well known machine learning models are also applied in various setup along with the feature engineering techniques to build predictive model for the disease prediction. The experiments show promising results.