Project investigating relationship between CKD and biological markers
The contents of this project include the R scripts, which perform the following tasks:
Transformation of the data-set
-
artificial joining of the data-sets (pre-split) and loading
-
investigating the number of NAs, replacing the numerical ones with median and dropping the categorical; including converting character type variables to numerical
Exploratory analysis
-
building a correlation heat map
-
investigating the relationship between correlated variables (distribution and frequency plots)
Modelling
-
linear regression for all predictors
-
linear regression for the most two correlated predictors
-
K-nearest neighbors modelling
Principal component analysis
Functions
- functions that help in building the ROC and confusion matrix plots