The goal of the project is to classify insurance applicants into the 8 different risk classes, based on anonymized features. Such features include family history, employment history, age, height, weight and BMI amongst others (See details of data in the data source section). The classification model will be evaluated based on quadratic weighted kappa. The quadratic weighted kappa measures the agreement between two ratings (ie. the actual vs predicted). The metric ranges from 0 (random agreement between raters) to 1 (complete agreement between raters).
code (comprise all the notebooks used, refer to the next section for detailed contents)
data (comprise datasets used, specifically the training data and test data)
assets (comprise visualizations saved)
a. [Data Cleaning](code/ b. c.
Figure x: sksdsdsds
Dataset is obtained from a Kaggle competition here.