Introduction

This report details the steps and results of a machine learning pipeline developed for cancer data analysis. It includes data preprocessing, various feature selection techniques (Near-Zero Variance and Boruta), hyperparameter tuning of a Random Forest model, class imbalance handling using SMOTE, and evaluation using 10-fold cross-validation and a hold-out test set. The impact of key steps like different feature selection methods and SMOTE is also visualized.

Configuration and Data Loading

Global Configuration

The following global configurations were used for the analysis:

file_path <- "Norm_avg_plasma_data - Sheet2.csv" # IMPORTANT: Ensure this file is in your R working directory
positive_class_name <- "Cancer"
negative_class_name <- "Normal"

Data Loading and Initial Inspection

The dataset is loaded from Norm_avg_plasma_data - Sheet2.csv. It’s assumed to have samples as rows, a header row, with the first column for sample identifiers, the second for class labels, and subsequent columns for feature values.

## Dimensions of raw data:  75  rows,  2813  columns.

## Column names from CSV header:

Raw Data Column Names (Transposed for Readability)
Unique.Id	Sample.type	X0	X0.1	X0.117	X0.133333	X0.15	X0.166667	X0.183	X0.2	X0.216667	X0.233333	X0.25	X0.266667	X0.283333	X0.3	X0.317	X0.333333	X0.35	X0.366667	X0.383	X0.4	X0.416667	X0.433333	X0.45	X0.466667	X0.483333	X0.5	X0.517	X0.533333	X0.55	X0.566667	X0.583	X0.6	X0.616667	X0.633333	X0.65	X0.666667	X0.683333	X0.7	X0.717	X0.733333	X0.75	X0.766667	X0.783	X0.8	X0.816667	X0.833333	X0.85	X0.867	X0.883333	X0.9	X0.916667	X0.933333	X0.95	X0.966667	X0.983	X1	X1.016667	X1.033333	X1.05	X1.066667	X1.083333	X1.1	X1.116667	X1.133333	X1.15	X1.166667	X1.183333	X1.2	X1.216667	X1.233333	X1.25	X1.266667	X1.283333	X1.3	X1.316667	X1.333333	X1.35	X1.366667	X1.383333	X1.4	X1.416667	X1.433333	X1.45	X1.466667	X1.483333	X1.5	X1.516667	X1.533333	X1.55	X1.566667	X1.583333	X1.6	X1.616667	X1.633333	X1.65	X1.666667	X1.683333	X1.7	X1.716667	X1.733333	X1.75	X1.766667	X1.783333	X1.8	X1.816667	X1.833333	X1.85	X1.866667	X1.883333	X1.9	X1.916667	X1.933333	X1.95	X1.966667	X1.983333	X2	X2.016667	X2.033333	X2.05	X2.066667	X2.083333	X2.1	X2.116667	X2.133333	X2.15	X2.166667	X2.183333	X2.2	X2.216667	X2.233333	X2.25	X2.266667	X2.283333	X2.3	X2.316667	X2.333333	X2.35	X2.366667	X2.383333	X2.4	X2.416667	X2.433333	X2.45	X2.466667	X2.483333	X2.5	X2.516667	X2.533333	X2.55	X2.566667	X2.583333	X2.6	X2.616667	X2.633333	X2.65	X2.666667	X2.683333	X2.7	X2.716667	X2.733333	X2.75	X2.766667	X2.783333	X2.8	X2.816667	X2.833333	X2.85	X2.866667	X2.883333	X2.9	X2.916667	X2.933333	X2.95	X2.966667	X2.983333	X3	X3.016667	X3.033333	X3.05	X3.066667	X3.083333	X3.1	X3.116667	X3.133333	X3.15	X3.166667	X3.183333	X3.2	X3.216667	X3.233333	X3.25	X3.266667	X3.283333	X3.3	X3.316667	X3.333333	X3.35	X3.366667	X3.383333	X3.4	X3.416667	X3.433333	X3.45	X3.466667	X3.483333	X3.5	X3.516667	X3.533333	X3.55	X3.566667	X3.583333	X3.6	X3.616667	X3.633333	X3.65	X3.666667	X3.683333	X3.7	X3.716667	X3.733333	X3.75	X3.766667	X3.783333	X3.8	X3.816667	X3.833333	X3.85	X3.866667	X3.883333	X3.9	X3.916667	X3.933333	X3.95	X3.966667	X3.983333	X4	X4.016667	X4.033333	X4.05	X4.066667	X4.083333	X4.1	X4.116667	X4.133333	X4.15	X4.166667	X4.183333	X4.2	X4.216667	X4.233333	X4.25	X4.266667	X4.283333	X4.3	X4.316667	X4.333333	X4.35	X4.366667	X4.383333	X4.4	X4.416667	X4.433333	X4.45	X4.466667	X4.483333	X4.5	X4.516667	X4.533333	X4.55	X4.566667	X4.583333	X4.6	X4.616667	X4.633333	X4.65	X4.666667	X4.683333	X4.7	X4.716667	X4.733333	X4.75	X4.766667	X4.783333	X4.8	X4.816667	X4.833333	X4.85	X4.866667	X4.883333	X4.9	X4.916667	X4.933333	X4.95	X4.966667	X4.983333	X5	X5.016667	X5.033333	X5.05	X5.066667	X5.083333	X5.1	X5.116667	X5.133333	X5.15	X5.166667	X5.183333	X5.2	X5.216667	X5.233333	X5.25	X5.266667	X5.283333	X5.3	X5.316667	X5.333333	X5.35	X5.366667	X5.383333	X5.4	X5.416667	X5.433333	X5.45	X5.466667	X5.483333	X5.5	X5.516667	X5.533333	X5.55	X5.566667	X5.583333	X5.6	X5.616667	X5.633333	X5.65	X5.666667	X5.683333	X5.7	X5.716667	X5.733333	X5.75	X5.766667	X5.783333	X5.8	X5.816667	X5.833333	X5.85	X5.866667	X5.883333	X5.9	X5.916667	X5.933333	X5.95	X5.966667	X5.983333	X6	X6.016667	X6.033333	X6.05	X6.066667	X6.083333	X6.1	X6.116667	X6.133333	X6.15	X6.166667	X6.183333	X6.2	X6.216667	X6.233333	X6.25	X6.266667	X6.283333	X6.3	X6.316667	X6.333333	X6.35	X6.366667	X6.383333	X6.4	X6.416667	X6.433333	X6.45	X6.466667	X6.483333	X6.5	X6.516667	X6.533333	X6.55	X6.566667	X6.583333	X6.6	X6.616667	X6.633333	X6.65	X6.666667	X6.683333	X6.7	X6.716667	X6.733333	X6.75	X6.766667	X6.783333	X6.8	X6.816667	X6.833333	X6.85	X6.866667	X6.883333	X6.9	X6.916667	X6.933333	X6.95	X6.966667	X6.983333	X7	X7.016667	X7.033333	X7.05	X7.066667	X7.083333	X7.1	X7.116667	X7.133333	X7.15	X7.166667	X7.183333	X7.2	X7.216667	X7.233333	X7.25	X7.266667	X7.283333	X7.3	X7.316667	X7.333333	X7.35	X7.366667	X7.383333	X7.4	X7.416667	X7.433333	X7.45	X7.466667	X7.483333	X7.5	X7.516667	X7.533333	X7.55	X7.566667	X7.583333	X7.6	X7.616667	X7.633333	X7.65	X7.666667	X7.683333	X7.7	X7.716667	X7.733333	X7.75	X7.766667	X7.783333	X7.8	X7.816667	X7.833333	X7.85	X7.866667	X7.883333	X7.9	X7.916667	X7.933333	X7.95	X7.966667	X7.983333	X8	X8.016667	X8.033333	X8.05	X8.066667	X8.083333	X8.1	X8.116667	X8.133333	X8.15	X8.166667	X8.183333	X8.2	X8.216667	X8.233333	X8.25	X8.266667	X8.283333	X8.3	X8.316667	X8.333333	X8.35	X8.366667	X8.383333	X8.4	X8.416667	X8.433333	X8.45	X8.466667	X8.483333	X8.5	X8.516667	X8.533333	X8.55	X8.566667	X8.583333	X8.6	X8.616667	X8.633333	X8.65	X8.666667	X8.683333	X8.7	X8.716667	X8.733333	X8.75	X8.766667	X8.783333	X8.8	X8.816667	X8.833333	X8.85	X8.866667	X8.883333	X8.9	X8.916667	X8.933333	X8.95	X8.966667	X8.983333	X9	X9.016667	X9.033333	X9.05	X9.066667	X9.083333	X9.1	X9.116667	X9.133333	X9.15	X9.166667	X9.183333	X9.2	X9.216667	X9.233333	X9.25	X9.266667	X9.283333	X9.3	X9.316667	X9.333333	X9.35	X9.366667	X9.383333	X9.4	X9.416667	X9.433333	X9.45	X9.466667	X9.483333	X9.5	X9.516667	X9.533333	X9.55	X9.566667	X9.583333	X9.6	X9.616667	X9.633333	X9.65	X9.666667	X9.683333	X9.7	X9.716667	X9.733333	X9.75	X9.766667	X9.783333	X9.8	X9.816667	X9.833333	X9.85	X9.866667	X9.883333	X9.9	X9.916667	X9.933333	X9.95	X9.966667	X9.983333	X10	X10.01667	X10.03333	X10.05	X10.06667	X10.08333	X10.1	X10.11667	X10.13333	X10.15	X10.16667	X10.18333	X10.2	X10.21667	X10.23333	X10.25	X10.26667	X10.28333	X10.3	X10.31667	X10.33333	X10.35	X10.36667	X10.38333	X10.4	X10.41667	X10.43333	X10.45	X10.46667	X10.48333	X10.5	X10.51667	X10.53333	X10.55	X10.56667	X10.58333	X10.6	X10.61667	X10.63333	X10.65	X10.66667	X10.68333	X10.7	X10.71667	X10.73333	X10.75	X10.76667	X10.78333	X10.8	X10.81667	X10.83333	X10.85	X10.86667	X10.88333	X10.9	X10.91667	X10.93333	X10.95	X10.96667	X10.98333	X11	X11.01667	X11.03333	X11.05	X11.06667	X11.08333	X11.1	X11.11667	X11.13333	X11.15	X11.16667	X11.18333	X11.2	X11.21667	X11.23333	X11.25	X11.26667	X11.28333	X11.3	X11.31667	X11.33333	X11.35	X11.36667	X11.38333	X11.4	X11.41667	X11.43333	X11.45	X11.46667	X11.48333	X11.5	X11.51667	X11.53333	X11.55	X11.56667	X11.58333	X11.6	X11.61667	X11.63333	X11.65	X11.66667	X11.68333	X11.7	X11.71667	X11.73333	X11.75	X11.76667	X11.78333	X11.8	X11.81667	X11.83333	X11.85	X11.86667	X11.88333	X11.9	X11.91667	X11.93333	X11.95	X11.96667	X11.98333	X12	X12.01667	X12.03333	X12.05	X12.06667	X12.08333	X12.1	X12.11667	X12.13333	X12.15	X12.16667	X12.18333	X12.2	X12.21667	X12.23333	X12.25	X12.26667	X12.28333	X12.3	X12.31667	X12.33333	X12.35	X12.36667	X12.38333	X12.4	X12.41667	X12.43333	X12.45	X12.46667	X12.48333	X12.5	X12.51667	X12.53333	X12.55	X12.56667	X12.58333	X12.6	X12.61667	X12.63333	X12.65	X12.66667	X12.68333	X12.7	X12.71667	X12.73333	X12.75	X12.76667	X12.78333	X12.8	X12.81667	X12.83333	X12.85	X12.86667	X12.88333	X12.9	X12.91667	X12.93333	X12.95	X12.96667	X12.98333	X13	X13.01667	X13.03333	X13.05	X13.06667	X13.08333	X13.1	X13.11667	X13.13333	X13.15	X13.16667	X13.18333	X13.2	X13.21667	X13.23333	X13.25	X13.26667	X13.28333	X13.3	X13.31667	X13.33333	X13.35	X13.36667	X13.38333	X13.4	X13.41667	X13.43333	X13.45	X13.46667	X13.48333	X13.5	X13.51667	X13.53333	X13.55	X13.56667	X13.58333	X13.6	X13.61667	X13.63333	X13.65	X13.66667	X13.68333	X13.7	X13.71667	X13.73333	X13.75	X13.76667	X13.78333	X13.8	X13.81667	X13.83333	X13.85	X13.86667	X13.88333	X13.9	X13.91667	X13.93333	X13.95	X13.96667	X13.98333	X14	X14.01667	X14.03333	X14.05	X14.06667	X14.08333	X14.1	X14.11667	X14.13333	X14.15	X14.16667	X14.18333	X14.2	X14.21667	X14.23333	X14.25	X14.26667	X14.28333	X14.3	X14.31667	X14.33333	X14.35	X14.36667	X14.38333	X14.4	X14.41667	X14.43333	X14.45	X14.46667	X14.48333	X14.5	X14.51667	X14.53333	X14.55	X14.56667	X14.58333	X14.6	X14.61667	X14.63333	X14.65	X14.66667	X14.68333	X14.7	X14.71667	X14.73333	X14.75	X14.76667	X14.78333	X14.8	X14.81667	X14.83333	X14.85	X14.86667	X14.88333	X14.9	X14.91667	X14.93333	X14.95	X14.96667	X14.98333	X15	X15.01667	X15.03333	X15.05	X15.06667	X15.08333	X15.1	X15.11667	X15.13333	X15.15	X15.16667	X15.18333	X15.2	X15.21667	X15.23333	X15.25	X15.26667	X15.28333	X15.3	X15.31667	X15.33333	X15.35	X15.36667	X15.38333	X15.4	X15.41667	X15.43333	X15.45	X15.46667	X15.48333	X15.5	X15.51667	X15.53333	X15.55	X15.56667	X15.58333	X15.6	X15.61667	X15.63333	X15.65	X15.66667	X15.68333	X15.7	X15.71667	X15.73333	X15.75	X15.76667	X15.78333	X15.8	X15.81667	X15.83333	X15.85	X15.86667	X15.88333	X15.9	X15.91667	X15.93333	X15.95	X15.96667	X15.98333	X16	X16.01667	X16.03333	X16.05	X16.06667	X16.08333	X16.1	X16.11667	X16.13333	X16.15	X16.16667	X16.18333	X16.2	X16.21667	X16.23333	X16.25	X16.26667	X16.28333	X16.3	X16.31667	X16.33333	X16.35	X16.36667	X16.38333	X16.4	X16.41667	X16.43333	X16.45	X16.46667	X16.48333	X16.5	X16.51667	X16.53333	X16.55	X16.56667	X16.58333	X16.6	X16.61667	X16.63333	X16.65	X16.66667	X16.68333	X16.7	X16.71667	X16.73333	X16.75	X16.76667	X16.78333	X16.8	X16.81667	X16.83333	X16.85	X16.86667	X16.88333	X16.9	X16.91667	X16.93333	X16.95	X16.96667	X16.98333	X17	X17.01667	X17.03333	X17.05	X17.06667	X17.08333	X17.1	X17.11667	X17.13333	X17.15	X17.16667	X17.18333	X17.2	X17.21667	X17.23333	X17.25	X17.26667	X17.28333	X17.3	X17.31667	X17.33333	X17.35	X17.36667	X17.38333	X17.4	X17.41667	X17.43333	X17.45	X17.46667	X17.48333	X17.5	X17.51667	X17.53333	X17.55	X17.56667	X17.58333	X17.6	X17.61667	X17.63333	X17.65	X17.66667	X17.68333	X17.7	X17.71667	X17.73333	X17.75	X17.76667	X17.78333	X17.8	X17.81667	X17.83333	X17.85	X17.86667	X17.88333	X17.9	X17.91667	X17.93333	X17.95	X17.96667	X17.98333	X18	X18.01667	X18.03333	X18.05	X18.06667	X18.08333	X18.1	X18.11667	X18.13333	X18.15	X18.16667	X18.18333	X18.2	X18.21667	X18.23333	X18.25	X18.26667	X18.28333	X18.3	X18.31667	X18.33333	X18.35	X18.36667	X18.38333	X18.4	X18.41667	X18.43333	X18.45	X18.46667	X18.48333	X18.5	X18.51667	X18.53333	X18.55	X18.56667	X18.58333	X18.6	X18.61667	X18.63333	X18.65	X18.66667	X18.68333	X18.7	X18.71667	X18.73333	X18.75	X18.76667	X18.78333	X18.8	X18.81667	X18.83333	X18.85	X18.86667	X18.88333	X18.9	X18.91667	X18.93333	X18.95	X18.96667	X18.98333	X19	X19.01667	X19.03333	X19.05	X19.06667	X19.08333	X19.1	X19.11667	X19.13333	X19.15	X19.16667	X19.18333	X19.2	X19.21667	X19.23333	X19.25	X19.26667	X19.28333	X19.3	X19.31667	X19.33333	X19.35	X19.36667	X19.38333	X19.4	X19.41667	X19.43333	X19.45	X19.46667	X19.48333	X19.5	X19.51667	X19.53333	X19.55	X19.56667	X19.58333	X19.6	X19.61667	X19.63333	X19.65	X19.66667	X19.68333	X19.7	X19.71667	X19.73333	X19.75	X19.76667	X19.78333	X19.8	X19.81667	X19.83333	X19.85	X19.86667	X19.88333	X19.9	X19.91667	X19.93333	X19.95	X19.96667	X19.98333	X20	X20.01667	X20.03333	X20.05	X20.06667	X20.08333	X20.1	X20.11667	X20.13333	X20.15	X20.16667	X20.18333	X20.2	X20.21667	X20.23333	X20.25	X20.26667	X20.28333	X20.3	X20.31667	X20.33333	X20.35	X20.36667	X20.38333	X20.4	X20.41667	X20.43333	X20.45	X20.46667	X20.48333	X20.5	X20.51667	X20.53333	X20.55	X20.56667	X20.58333	X20.6	X20.61667	X20.63333	X20.65	X20.66667	X20.68333	X20.7	X20.71667	X20.73333	X20.75	X20.76667	X20.78333	X20.8	X20.81667	X20.83333	X20.85	X20.86667	X20.88333	X20.9	X20.91667	X20.93333	X20.95	X20.96667	X20.98333	X21	X21.01667	X21.03333	X21.05	X21.06667	X21.08333	X21.1	X21.11667	X21.13333	X21.15	X21.16667	X21.18333	X21.2	X21.21667	X21.23333	X21.25	X21.26667	X21.28333	X21.3	X21.31667	X21.33333	X21.35	X21.36667	X21.38333	X21.4	X21.41667	X21.43333	X21.45	X21.46667	X21.48333	X21.5	X21.51667	X21.53333	X21.55	X21.56667	X21.58333	X21.6	X21.61667	X21.63333	X21.65	X21.66667	X21.68333	X21.7	X21.71667	X21.73333	X21.75	X21.76667	X21.78333	X21.8	X21.81667	X21.83333	X21.85	X21.86667	X21.88333	X21.9	X21.91667	X21.93333	X21.95	X21.96667	X21.98333	X22	X22.01667	X22.03333	X22.05	X22.06667	X22.08333	X22.1	X22.11667	X22.13333	X22.15	X22.16667	X22.18333	X22.2	X22.21667	X22.23333	X22.25	X22.26667	X22.28333	X22.3	X22.31667	X22.33333	X22.35	X22.36667	X22.38333	X22.4	X22.41667	X22.43333	X22.45	X22.46667	X22.48333	X22.5	X22.51667	X22.53333	X22.55	X22.56667	X22.58333	X22.6	X22.61667	X22.63333	X22.65	X22.66667	X22.68333	X22.7	X22.71667	X22.73333	X22.75	X22.76667	X22.78333	X22.8	X22.81667	X22.83333	X22.85	X22.86667	X22.88333	X22.9	X22.91667	X22.93333	X22.95	X22.96667	X22.98333	X23	X23.01667	X23.03333	X23.05	X23.06667	X23.08333	X23.1	X23.11667	X23.13333	X23.15	X23.16667	X23.18333	X23.2	X23.21667	X23.23333	X23.25	X23.26667	X23.28333	X23.3	X23.31667	X23.33333	X23.35	X23.36667	X23.38333	X23.4	X23.41667	X23.43333	X23.45	X23.46667	X23.48333	X23.5	X23.51667	X23.53333	X23.55	X23.56667	X23.58333	X23.6	X23.61667	X23.63333	X23.65	X23.66667	X23.68333	X23.7	X23.71667	X23.73333	X23.75	X23.76667	X23.78333	X23.8	X23.81667	X23.83333	X23.85	X23.86667	X23.88333	X23.9	X23.91667	X23.93333	X23.95	X23.96667	X23.98333	X24	X24.01667	X24.03333	X24.05	X24.06667	X24.08333	X24.1	X24.11667	X24.13333	X24.15	X24.16667	X24.18333	X24.2	X24.21667	X24.23333	X24.25	X24.26667	X24.28333	X24.3	X24.31667	X24.33333	X24.35	X24.36667	X24.38333	X24.4	X24.41667	X24.43333	X24.45	X24.46667	X24.48333	X24.5	X24.51667	X24.53333	X24.55	X24.56667	X24.58333	X24.6	X24.61667	X24.63333	X24.65	X24.66667	X24.68333	X24.7	X24.71667	X24.73333	X24.75	X24.76667	X24.78333	X24.8	X24.81667	X24.83333	X24.85	X24.86667	X24.88333	X24.9	X24.91667	X24.93333	X24.95	X24.96667	X24.98333	X25	X25.01667	X25.03333	X25.05	X25.06667	X25.08333	X25.1	X25.11667	X25.13333	X25.15	X25.16667	X25.18333	X25.2	X25.21667	X25.23333	X25.25	X25.26667	X25.28333	X25.3	X25.31667	X25.33333	X25.35	X25.36667	X25.38333	X25.4	X25.41667	X25.43333	X25.45	X25.46667	X25.48333	X25.5	X25.51667	X25.53333	X25.55	X25.56667	X25.58333	X25.6	X25.61667	X25.63333	X25.65	X25.66667	X25.68333	X25.7	X25.71667	X25.73333	X25.75	X25.76667	X25.78333	X25.8	X25.81667	X25.83333	X25.85	X25.86667	X25.88333	X25.9	X25.91667	X25.93333	X25.95	X25.96667	X25.98333	X26	X26.01667	X26.03333	X26.05	X26.06667	X26.08333	X26.1	X26.11667	X26.13333	X26.15	X26.16667	X26.18333	X26.2	X26.21667	X26.23333	X26.25	X26.26667	X26.28333	X26.3	X26.31667	X26.33333	X26.35	X26.36667	X26.38333	X26.4	X26.41667	X26.43333	X26.45	X26.46667	X26.48333	X26.5	X26.51667	X26.53333	X26.55	X26.56667	X26.58333	X26.6	X26.61667	X26.63333	X26.65	X26.66667	X26.68333	X26.7	X26.71667	X26.73333	X26.75	X26.76667	X26.78333	X26.8	X26.81667	X26.83333	X26.85	X26.86667	X26.88333	X26.9	X26.91667	X26.93333	X26.95	X26.96667	X26.98333	X27	X27.01667	X27.03333	X27.05	X27.06667	X27.08333	X27.1	X27.11667	X27.13333	X27.15	X27.16667	X27.18333	X27.2	X27.21667	X27.23333	X27.25	X27.26667	X27.28333	X27.3	X27.31667	X27.33333	X27.35	X27.36667	X27.38333	X27.4	X27.41667	X27.43333	X27.45	X27.46667	X27.48333	X27.5	X27.51667	X27.53333	X27.55	X27.56667	X27.58333	X27.6	X27.61667	X27.63333	X27.65	X27.66667	X27.68333	X27.7	X27.71667	X27.73333	X27.75	X27.76667	X27.78333	X27.8	X27.81667	X27.83333	X27.85	X27.86667	X27.88333	X27.9	X27.91667	X27.93333	X27.95	X27.96667	X27.98333	X28	X28.01667	X28.03333	X28.05	X28.06667	X28.08333	X28.1	X28.11667	X28.13333	X28.15	X28.16667	X28.18333	X28.2	X28.21667	X28.23333	X28.25	X28.26667	X28.28333	X28.3	X28.31667	X28.33333	X28.35	X28.36667	X28.38333	X28.4	X28.41667	X28.43333	X28.45	X28.46667	X28.48333	X28.5	X28.51667	X28.53333	X28.55	X28.56667	X28.58333	X28.6	X28.61667	X28.63333	X28.65	X28.66667	X28.68333	X28.7	X28.71667	X28.73333	X28.75	X28.76667	X28.78333	X28.8	X28.81667	X28.83333	X28.85	X28.86667	X28.88333	X28.9	X28.91667	X28.93333	X28.95	X28.96667	X28.98333	X29	X29.01667	X29.03333	X29.05	X29.06667	X29.08333	X29.1	X29.11667	X29.13333	X29.15	X29.16667	X29.18333	X29.2	X29.21667	X29.23333	X29.25	X29.26667	X29.28333	X29.3	X29.31667	X29.33333	X29.35	X29.36667	X29.38333	X29.4	X29.41667	X29.43333	X29.45	X29.46667	X29.48333	X29.5	X29.51667	X29.53333	X29.55	X29.56667	X29.58333	X29.6	X29.61667	X29.63333	X29.65	X29.66667	X29.68333	X29.7	X29.71667	X29.73333	X29.75	X29.76667	X29.78333	X29.8	X29.81667	X29.83333	X29.85	X29.86667	X29.88333	X29.9	X29.91667	X29.93333	X29.95	X29.96667	X29.98333	X30	X30.01667	X30.03333	X30.05	X30.06667	X30.08333	X30.1	X30.11667	X30.13333	X30.15	X30.16667	X30.18333	X30.2	X30.21667	X30.23333	X30.25	X30.26667	X30.28333	X30.3	X30.31667	X30.33333	X30.35	X30.36667	X30.38333	X30.4	X30.41667	X30.43333	X30.45	X30.46667	X30.48333	X30.5	X30.51667	X30.53333	X30.55	X30.56667	X30.58333	X30.6	X30.61667	X30.63333	X30.65	X30.66667	X30.68333	X30.7	X30.71667	X30.73333	X30.75	X30.76667	X30.78333	X30.8	X30.81667	X30.83333	X30.85	X30.86667	X30.88333	X30.9	X30.91667	X30.93333	X30.95	X30.96667	X30.98333	X31	X31.01667	X31.03333	X31.05	X31.06667	X31.08333	X31.1	X31.11667	X31.13333	X31.15	X31.16667	X31.18333	X31.2	X31.21667	X31.23333	X31.25	X31.26667	X31.28333	X31.3	X31.31667	X31.33333	X31.35	X31.36667	X31.38333	X31.4	X31.41667	X31.43333	X31.45	X31.46667	X31.48333	X31.5	X31.51667	X31.53333	X31.55	X31.56667	X31.58333	X31.6	X31.61667	X31.63333	X31.65	X31.66667	X31.68333	X31.7	X31.71667	X31.73333	X31.75	X31.76667	X31.78333	X31.8	X31.81667	X31.83333	X31.85	X31.86667	X31.88333	X31.9	X31.91667	X31.93333	X31.95	X31.96667	X31.98333	X32	X32.01667	X32.03333	X32.05	X32.06667	X32.08333	X32.1	X32.11667	X32.13333	X32.15	X32.16667	X32.18333	X32.2	X32.21667	X32.23333	X32.25	X32.26667	X32.28333	X32.3	X32.31667	X32.33333	X32.35	X32.36667	X32.38333	X32.4	X32.41667	X32.43333	X32.45	X32.46667	X32.48333	X32.5	X32.51667	X32.53333	X32.55	X32.56667	X32.58333	X32.6	X32.61667	X32.63333	X32.65	X32.66667	X32.68333	X32.7	X32.71667	X32.73333	X32.75	X32.76667	X32.78333	X32.8	X32.81667	X32.83333	X32.85	X32.86667	X32.88333	X32.9	X32.91667	X32.93333	X32.95	X32.96667	X32.98333	X33	X33.01667	X33.03333	X33.05	X33.06667	X33.08333	X33.1	X33.11667	X33.13333	X33.15	X33.16667	X33.18333	X33.2	X33.21667	X33.23333	X33.25	X33.26667	X33.28333	X33.3	X33.31667	X33.33333	X33.35	X33.36667	X33.38333	X33.4	X33.41667	X33.43333	X33.45	X33.46667	X33.48333	X33.5	X33.51667	X33.53333	X33.55	X33.56667	X33.58333	X33.6	X33.61667	X33.63333	X33.65	X33.66667	X33.68333	X33.7	X33.71667	X33.73333	X33.75	X33.76667	X33.78333	X33.8	X33.81667	X33.83333	X33.85	X33.86667	X33.88333	X33.9	X33.91667	X33.93333	X33.95	X33.96667	X33.98333	X34	X34.01667	X34.03333	X34.05	X34.06667	X34.08333	X34.1	X34.11667	X34.13333	X34.15	X34.16667	X34.18333	X34.2	X34.21667	X34.23333	X34.25	X34.26667	X34.28333	X34.3	X34.31667	X34.33333	X34.35	X34.36667	X34.38333	X34.4	X34.41667	X34.43333	X34.45	X34.46667	X34.48333	X34.5	X34.51667	X34.53333	X34.55	X34.56667	X34.58333	X34.6	X34.61667	X34.63333	X34.65	X34.66667	X34.68333	X34.7	X34.71667	X34.73333	X34.75	X34.76667	X34.78333	X34.8	X34.81667	X34.83333	X34.85	X34.86667	X34.88333	X34.9	X34.91667	X34.93333	X34.95	X34.96667	X34.98333	X35	X35.01667	Wavelength_200	X201	X202	X203	X204	X205	X206	X207	X208	X209	X210	X211	X212	X213	X214	X215	X216	X217	X218	X219	X220	X221	X222	X223	X224	X225	X226	X227	X228	X229	X230	X231	X232	X233	X234	X235	X236	X237	X238	X239	X240	X241	X242	X243	X244	X245	X246	X247	X248	X249	X250	X251	X252	X253	X254	X255	X256	X257	X258	X259	X260	X261	X262	X263	X264	X265	X266	X267	X268	X269	X270	X271	X272	X273	X274	X275	X276	X277	X278	X279	X280	X281	X282	X283	X284	X285	X286	X287	X288	X289	X290	X291	X292	X293	X294	X295	X296	X297	X298	X299	X300	X301	X302	X303	X304	X305	X306	X307	X308	X309	X310	X311	X312	X313	X314	X315	X316	X317	X318	X319	X320	X321	X322	X323	X324	X325	X326	X327	X328	X329	X330	X331	X332	X333	X334	X335	X336	X337	X338	X339	X340	X341	X342	X343	X344	X345	X346	X347	X348	X349	X350	X351	X352	X353	X354	X355	X356	X357	X358	X359	X360	X361	X362	X363	X364	X365	X366	X367	X368	X369	X370	X371	X372	X373	X374	X375	X376	X377	X378	X379	X380	X381	X382	X383	X384	X385	X386	X387	X388	X389	X390	X391	X392	X393	X394	X395	X396	X397	X398	X399	X400	X401	X402	X403	X404	X405	X406	X407	X408	X409	X410	X411	X412	X413	X414	X415	X416	X417	X418	X419	X420	X421	X422	X423	X424	X425	X426	X427	X428	X429	X430	X431	X432	X433	X434	X435	X436	X437	X438	X439	X440	X441	X442	X443	X444	X445	X446	X447	X448	X449	X450	X451	X452	X453	X454	X455	X456	X457	X458	X459	X460	X461	X462	X463	X464	X465	X466	X467	X468	X469	X470	X471	X472	X473	X474	X475	X476	X477	X478	X479	X480	X481	X482	X483	X484	X485	X486	X487	X488	X489	X490	X491	X492	X493	X494	X495	X496	X497	X498	X499	X500	X501	X502	X503	X504	X505	X506	X507	X508	X509	X510	X511	X512	X513	X514	X515	X516	X517	X518	X519	X520	X521	X522	X523	X524	X525	X526	X527	X528	X529	X530	X531	X532	X533	X534	X535	X536	X537	X538	X539	X540	X541	X542	X543	X544	X545	X546	X547	X548	X549	X550	X551	X552	X553	X554	X555	X556	X557	X558	X559	X560	X561	X562	X563	X564	X565	X566	X567	X568	X569	X570	X571	X572	X573	X574	X575	X576	X577	X578	X579	X580	X581	X582	X583	X584	X585	X586	X587	X588	X589	X590	X591	X592	X593	X594	X595	X596	X597	X598	X599	X600	X601	X602	X603	X604	X605	X606	X607	X608	X609	X610	X611	X612	X613	X614	X615	X616	X617	X618	X619	X620	X621	X622	X623	X624	X625	X626	X627	X628	X629	X630	X631	X632	X633	X634	X635	X636	X637	X638	X639	X640	X641	X642	X643	X644	X645	X646	X647	X648	X649	X650	X651	X652	X653	X654	X655	X656	X657	X658	X659	X660	X661	X662	X663	X664	X665	X666	X667	X668	X669	X670	X671	X672	X673	X674	X675	X676	X677	X678	X679	X680	X681	X682	X683	X684	X685	X686	X687	X688	X689	X690	X691	X692	X693	X694	X695	X696	X697	X698	X699	X700	X701	X702	X703	X704	X705	X706	X707	X708	X709	X710	X711	X712	X713	X714	X715	X716	X717	X718	X719	X720	X721	X722	X723	X724	X725	X726	X727	X728	X729	X730	X731	X732	X733	X734	X735	X736	X737	X738	X739	X740	X741	X742	X743	X744	X745	X746	X747	X748	X749	X750	X751	X752	X753	X754	X755	X756	X757	X758	X759	X760	X761	X762	X763	X764	X765	X766	X767	X768	X769	X770	X771	X772	X773	X774	X775	X776	X777	X778	X779	X780	X781	X782	X783	X784	X785	X786	X787	X788	X789	X790	X791	X792	X793	X794	X795	X796	X797	X798	X799	X800	X801	X802	X803	X804	X805	X806	X807	X808	X809	X810	X811	X812	X813	X814	X815	X816	X817	X818	X819	X820	X821	X822	X823	X824	X825	X826	X827	X828	X829	X830	X831	X832	X833	X834	X835	X836	X837	X838	X839	X840	X841	X842	X843	X844	X845	X846	X847	X848	X849	X850	X851	X852	X853	X854	X855	X856	X857	X858	X859	X860	X861	X862	X863	X864	X865	X866	X867	X868	X869	X870	X871	X872	X873	X874	X875	X876	X877	X878	X879	X880	X881	X882	X883	X884	X885	X886	X887	X888	X889	X890	X891	X892	X893	X894	X895	X896	X897	X898	X899	X900	X1.10	X2.10	X3.10	X4.10	X5.10	X6.10	X7.10	X8.10	X9.10	X10.10	X11.10	X12.10	X13.10

## 
## First 5 rows and 5 columns of raw data:

Head of Raw Data
Unique.Id	Sample.type	X0	X0.1	X0.117
PC1	Cancer	-3.63e-03	-0.002010	-1.52e-03
PC2	Cancer	-7.33e-05	-0.000210	-3.70e-04
PC3	Cancer	-1.47e-05	0.000195	1.75e-04
PC4	Cancer	-8.60e-04	0.000411	7.00e-04
PC5	Cancer	1.12e-04	0.000057	1.93e-05

Initial Data Preparation

This section covers the initial processing of the raw data to prepare it for feature engineering and modeling. This includes extracting sample IDs, class labels, and features, converting features to numeric, and handling any samples with missing class labels. This prepared dataset (df_initial_features) will be the input for feature selection methods.

## Number of features identified initially:  2811

## Dimensions of initial feature data (df_initial_features) after NA Class removal:  75  samples,  2812  columns (including Class).

## Class distribution in df_initial_features:

Class Distribution in Initial Data (for Feature Selection)
Var1	Freq
Normal	48
Cancer	27

Feature Selection Methods

We explore two feature selection methods: Near-Zero Variance (NZV) filtering and Boruta.

Feature Reduction: Near-Zero Variance (NZV)

NZV filtering removes features with little to no variance.

## Number of near-zero variance features removed:  0

## Number of features remaining after NZV filtering:  2811

## Dimensions of NZV-filtered feature data (df_nzv_features):  75  samples,  2812  columns (including Class).

Feature Selection: Boruta

Boruta is a wrapper algorithm built around Random Forest that iteratively compares the importance of original features with that of random shadow features.

Note: Boruta can be computationally intensive, especially with a large number of features or samples. The maxRuns parameter controls the maximum number of Random Forest runs.

## Starting Boruta feature selection... This may take some time.

## Boruta feature selection complete.

## Number of features selected by Boruta (Confirmed only):  7

## Dimensions of Boruta-selected feature data (df_boruta_features):  75  samples,  8  columns (including Class).

Model Training Framework

Helper Functions

# Helper function for training and preprocessing
train_rf_model <- function(data_to_train, model_name_suffix, train_control_config, tune_grid_config) {
    set.seed(123) 
    train_idx_helper <- createDataPartition(data_to_train$Class, p = .80, list = FALSE, times = 1)
    cv_train_data_helper <- data_to_train[train_idx_helper, ]
    
    original_full_data_rownames <- rownames(data_to_train)
    holdout_indices_from_original <- which(!original_full_data_rownames %in% rownames(cv_train_data_helper))

    cat(paste("\n--- Training RF Model:", model_name_suffix, "---\n"))
    cat("Dimensions of CV training data for", model_name_suffix, ":", dim(cv_train_data_helper)[1],"x",dim(cv_train_data_helper)[2],"\n")
    
    features_for_preproc_helper <- cv_train_data_helper[, -which(names(cv_train_data_helper) == "Class"), drop = FALSE]
    # Ensure there are features to preprocess
    if(ncol(features_for_preproc_helper) == 0) {
        cat("Warning: No features to preprocess for model", model_name_suffix, ". Model training might fail or be trivial.\n")
        # Create a dummy processed_cv_train_data if no features, to allow train to proceed (it will likely be a majority class classifier)
        processed_cv_train_data_helper <- cv_train_data_helper[, "Class", drop=FALSE]
        current_preProcValues_helper <- NULL # No preprocessor
    } else {
        current_preProcValues_helper <- preProcess(features_for_preproc_helper, method = c("center", "scale", "medianImpute"))
        processed_cv_train_features_helper <- predict(current_preProcValues_helper, features_for_preproc_helper)
        processed_cv_train_data_helper <- cbind(processed_cv_train_features_helper, Class = cv_train_data_helper$Class)
    }

    minority_size_helper <- min(table(processed_cv_train_data_helper$Class))
    if ("sampling" %in% names(train_control_config) && !is.null(train_control_config$sampling) && minority_size_helper < 10 && ncol(features_for_preproc_helper) > 0) {
        # Warning for SMOTE
    }

    set.seed(456)
    if(ncol(processed_cv_train_data_helper) <= 1 && !"Class" %in% colnames(processed_cv_train_data_helper)) { # Only class or empty
        cat("Skipping model training for", model_name_suffix, "due to no predictive features after preprocessing.\n")
        return(list(model = NULL, preprocessor = current_preProcValues_helper, holdout_indices_original = holdout_indices_from_original))
    }

    model_fit <- tryCatch({
        train(
            Class ~ .,
            data = processed_cv_train_data_helper,
            method = "rf",
            trControl = train_control_config,
            metric = "ROC",
            tuneGrid = tune_grid_config,
            importance = TRUE,
            na.action = na.omit
        )
    }, error = function(e) {
        cat("Error during training for", model_name_suffix, ":", e$message, "\n")
        return(NULL) # Return NULL if training fails
    })
    
    if(!is.null(model_fit)){
        cat("Training complete for RF Model:", model_name_suffix, "\n")
    }
    
    return(list(model = model_fit, 
                preprocessor = current_preProcValues_helper, 
                holdout_indices_original = holdout_indices_from_original))
}

create_rf_tune_grid <- function(data_frame_for_features) {
    # Check if there are any feature columns other than 'Class'
    feature_cols <- setdiff(colnames(data_frame_for_features), "Class")
    if(length(feature_cols) == 0) {
        # No features to tune mtry for, return a default grid that RF can handle (e.g. mtry=1 if Class is the only col, though RF will fail)
        # Or, more robustly, this case should be handled before calling train (e.g. skip training)
        # For now, let's assume if this function is called, there's at least one feature.
        # If features_for_preproc_helper is empty, this will lead to num_feats = 0
        warning("create_rf_tune_grid called with no feature columns. mtry grid will be minimal.")
        return(expand.grid(mtry = 1)) # Default mtry for RF if p=0 is problematic
    }
    num_feats <- length(feature_cols)
    default_m <- floor(sqrt(num_feats))
    grid_vals <- unique(c(
        max(1, floor(default_m / 2)), # ensure mtry is at least 1
        max(1, default_m),
        min(num_feats, max(1, default_m * 2)), 
        min(num_feats, 10), 
        min(num_feats, 20)
    ))
    grid_vals <- sort(unique(grid_vals[grid_vals <= num_feats & grid_vals > 0]))
    if(length(grid_vals) == 0) { grid_vals <- c(max(1, num_feats)) }
    return(expand.grid(mtry = grid_vals))
}

Base `trainControl` Configuration

base_train_control <- trainControl(
    method = "cv", number = 10, summaryFunction = twoClassSummary,
    classProbs = TRUE, verboseIter = FALSE, allowParallel = TRUE 
)

Model Training Scenarios and Comparative Analysis

Four Random Forest model scenarios are trained and compared: 1. All Features + SMOTE: Uses all initial features with SMOTE. 2. NZV Features + SMOTE: Uses NZV-filtered features with SMOTE. 3. Boruta Features + SMOTE (New Main Model): Uses Boruta-selected features with SMOTE. This is now considered the primary model. 4. Boruta Features + No SMOTE: Uses Boruta-selected features without SMOTE (for SMOTE impact on best feature set).

Scenario 1: All Features + SMOTE Tune grid (mtry values): 10, 20, 26, 53, 106

— Training RF Model: AllFeatures_SMOTE — Dimensions of CV training data for AllFeatures_SMOTE : 61 x 2812 Training complete for RF Model: AllFeatures_SMOTE Best mtry: 20 CV AUCROC (best tune): 0.8062

Scenario 2: NZV Features + SMOTE Tune grid (mtry values): 10, 20, 26, 53, 106

— Training RF Model: NZVFeatures_SMOTE — Dimensions of CV training data for NZVFeatures_SMOTE : 61 x 2812 Training complete for RF Model: NZVFeatures_SMOTE Best mtry: 20 CV AUCROC (best tune): 0.8062

Scenario 3: Boruta Features + SMOTE (Main Model) Tune grid (mtry values): 1, 2, 4, 7

— Training RF Model: BorutaFeatures_SMOTE (Main) — Dimensions of CV training data for BorutaFeatures_SMOTE (Main) : 61 x 8 Training complete for RF Model: BorutaFeatures_SMOTE (Main) Best mtry: 4 CV AUCROC (best tune): 0.7375

Scenario 4: Boruta Features + NO SMOTE Tune grid (mtry values): 1, 2, 4, 7

— Training RF Model: BorutaFeatures_NoSMOTE — Dimensions of CV training data for BorutaFeatures_NoSMOTE : 61 x 8 Training complete for RF Model: BorutaFeatures_NoSMOTE Best mtry: 1 CV AUCROC (best tune): 0.7458

Visualization of Key Step Impacts

Hyperparameter Tuning Profile (Main Boruta Model)

Hyperparameter Tuning Profile for the Main Model (Boruta Features + SMOTE).

Impact of Feature Selection Method

This plot compares the best cross-validated AUCROC from models trained with All Features, NZV-filtered Features, and Boruta-selected Features (all with SMOTE and hyperparameter tuning).

Impact of Feature Selection Method on Cross-Validated AUCROC.

Impact of SMOTE (on Boruta-Selected Features)

Impact of SMOTE on Performance (Boruta Features, Best Tune).

Main Model (Boruta Features + SMOTE): CV Performance Distribution

CV Performance Distribution (Main Boruta Model).

Main Model (Boruta Features + SMOTE): Evaluation on Hold-out Test Set

The main model (Boruta Features + SMOTE, best mtry) is evaluated on a separate hold-out test set.

## **Confusion Matrix (Hold-out - Main Boruta Model):**
## 
## **Overall Statistics (Hold-out - Main Boruta Model):**
## 
## **Class Statistics (Hold-out - Main Boruta Model):**
## 
## AUC (Hold-out - Main Boruta Model): 0.9333

ROC Curve for Main Boruta Model on Hold-out Test Set.

Main Model (Boruta Features + SMOTE): Variable Importance

The plot below shows the top features identified by the main model (Boruta Features + SMOTE, best mtry).

Top Important Features (Main Boruta Model).

Conclusion

This report presented a comprehensive machine learning pipeline, including data preprocessing, feature selection (NZV and Boruta), hyperparameter tuning for a Random Forest model, class imbalance handling (SMOTE), and thorough evaluation using 10-fold cross-validation and a hold-out test set. The main model utilized features selected by Boruta, along with SMOTE and hyperparameter tuning.

Key findings include: - The performance of models with different mtry values. - The impact of different feature selection methods (All Features vs. NZV vs. Boruta) on model performance (AUCROC). - The effect of SMOTE on AUCROC, Sensitivity, and Specificity for the Boruta-selected feature set. - The stability of the main Boruta model’s performance across 10 CV folds. - The final performance of the main Boruta model on an unseen hold-out test set. - The top features contributing to the main Boruta model’s predictions.

These results provide a solid foundation for understanding the dataset and the model’s predictive capabilities. Boruta feature selection, in conjunction with other pipeline steps, aimed to identify a robust and informative set of features for classification.

Machine Learning Pipeline Report for Cancer Data Analysis

Automated Report

2025-05-17

Introduction

Configuration and Data Loading

Global Configuration

Data Loading and Initial Inspection

Initial Data Preparation

Feature Selection Methods

Feature Reduction: Near-Zero Variance (NZV)

Feature Selection: Boruta

Model Training Framework

Helper Functions

Base `trainControl` Configuration

Model Training Scenarios and Comparative Analysis

Visualization of Key Step Impacts

Hyperparameter Tuning Profile (Main Boruta Model)

Impact of Feature Selection Method

Impact of SMOTE (on Boruta-Selected Features)

Main Model (Boruta Features + SMOTE): CV Performance Distribution

Main Model (Boruta Features + SMOTE): Evaluation on Hold-out Test Set

Main Model (Boruta Features + SMOTE): Variable Importance

Conclusion

Machine Learning Pipeline Report for Cancer Data Analysis

Automated Report

2025-05-17

Introduction

Configuration and Data Loading

Global Configuration

Data Loading and Initial Inspection

Initial Data Preparation

Feature Selection Methods

Feature Reduction: Near-Zero Variance (NZV)

Feature Selection: Boruta

Model Training Framework

Helper Functions

Base trainControl Configuration

Model Training Scenarios and Comparative Analysis

Visualization of Key Step Impacts

Hyperparameter Tuning Profile (Main Boruta Model)

Impact of Feature Selection Method

Impact of SMOTE (on Boruta-Selected Features)

Main Model (Boruta Features + SMOTE): CV Performance Distribution

Main Model (Boruta Features + SMOTE): Evaluation on Hold-out Test Set

Main Model (Boruta Features + SMOTE): Variable Importance

Conclusion

Base `trainControl` Configuration