Homework answers / question archive / Construct predictive models to examine determinants of medical costs billed by health insurance companies

Construct predictive models to examine determinants of medical costs billed by health insurance companies

Unrecognized

Share With

Construct predictive models to examine determinants of medical costs billed by health insurance companies. This information will be useful to eventually determine premiums. Below are the data variables: • Age: insurance contractor age, years • Sex: insurance contractor gender, [female, male] • BMI: Body mass index, providing an understanding of body, weights that are relatively high or low relative to height, objective index of body weight (kg / m ^ 2) using the ratio of height to weight, ideally 18.5 to 24.9 • Children: number of children covered by health insurance / Number of dependents • Smoker: smoking, [yes, no] • Region: the beneficiary’s residential area in the US, [northeast, southeast, southwest, northwest] • Charges: Individual medical costs billed by health insurance, $ (predicted value) The models we will consider are: - Boosted NN (NTanH(2), 500 boosts ) - Boosted NN (NTanH(2), robust, 500 boosts )- Note: robust option is similar to using Cauchy distribution in penalized regressions, it uses an estimation method (Laplacian likelihood function) that can produce better predictions if the outliers are important in the data set. - Boosted NN (NTanH(2), 500 boosts, learning rate=.05)- Note: this applies a smaller learning rate at each boosting, and hence, may take a bit longer, but potentially could improve results. - OLS as the base method -with all NN models, as well as the validation column as usual, use the random seed 123. identify most important variables. As a new case, suppose you have 45 year old non-smoker male with a BMI of 38 from the southeast, who has 2 children. What is your predicted medical costs for him? FOLLOW BELOW 4-5 pages long documents (12 points font, and 1.5 line spacing) Your paper should address the following points: 1) Introduction (20 points) - Describe the problem/question - Discuss the data set; response variable of interest and predictors, or any points you would like to mention - Mention the statistical methods you will employ 2) Analysis and Model Comparison (40 points) - Briefly discuss features (advantages/disadvantages, other specifications) of the methods you will employ - Describe the cross-validation procedure that you employ - Briefly outline the analysis you conducted - Report the Model Comparison results and the method you choose 3) Interpretation (40 points) - If a linear model is chosen, report parameter estimates and discuss their implications - Conduct variable importance analysis for your chosen model - Investigate your model’s profilers and interpret them - If the assignment includes a part to make a specific forecast, report your model’s prediction.