Data-Mining - Data Mining Assignment - 1 - Solved

Task I: Data Preprocessing
Suppose that you want to build a model to predict whether a client has a health coverage insurance. You collected a set of information that you think can help predict the probability of health coverage.

1-  Load the “Clients” dataset into R. What does the collected information on each client consist of?

2-  Using the summary function, comment on age, income, and housingStatus.

3-  Based on a subset of the data with 0 < 𝑎𝑔𝑒 < 100 and 𝑖𝑛𝑐𝑜𝑚𝑒 > 0, is there a correlation between age and income? Plot both variables against each other and comment on the variations.

4-  What is the number of missing values in housingStatus, recentlyChangedHousing, and numberCars variables? Find out whether those missing values come from the same observations? Comment on how to deal with them accordingly.

5-  Now check the variable works. What can you say about the number of missing values? Does it make sense to remove all observations with NAs? If you think about the possible meaning of NA for this variable, it seems more reasonable to create a new variable fixedWorks with an additional level of value “missing” for those observations, while keeping “employed” and “not employed” for the two other possible levels. Find a way to define such variable in R – Hint:  use ifelse.

6-  What is the type of variable income? What is the number of missing values in this variable? Assuming that the observations with missing income values have the same distribution as clients with specified income values, create a new variable fixedIncome in which you fill the mean income value in place of missing income values – Hint:  you can also use ifelse.

7-  Recall that you want to use this dataset to predict whether a client has a health coverage insurance based on collected information about him/her. Do you think that the actual value of age matter? Or is it the ranges of corresponding values that need to be considered to predict health coverage? In this direction, it may seem reasonable to discretize such variables, i.e. convert them from continuous to discrete variables.  

Use the function cut to define a new variable age.range by dividing age into ranges, with breaks 0,25,65,𝑎𝑛𝑑 𝐼𝑁𝐹.

 

 

 

Task II: The goal is to generate a model to predict diamond prices based on a set of features.

1-  Use ggplot to:

a.       Plot price against carat.

b.       Plot cut against price.

2-  Generate a simple linear regression model with price as response and carat as predictor. Comment on the model summary.  

3-  Generate a multiple linear regression model with price as response and carat, clarity, and color as predictors. Comment on the model summary.

4-  Generate a multiple linear regression model with price as response and all attribute in the dataset as predictors. Comment on the model summary and compare it to the previous model.