Parameters of attribute selection

The table below shows the parameters that were selected in the data selection process that resulted in the data file used to generate this report.

Data set

  • Number of columns: 9

  • Including numerical: 6

  • Including binary: 2

  • Number of rows: 109

  • Percent of missing values 0.82%

  • Number of variables with only 1 unique value: 0

  • Number of columns for which all values are missing: 0

  • Number of attributes taken into account in the analysis: 8

  • Decision class: outcome

  • Class positive value: dead

  • Attribute indicating survival status: outcome

  • Positive value for column with survival status: dead

  • Attribute indicating survival time: survival_time

Data source statistics:

  • Number of columns with methylation data: 4

  • Number of columns with RNA-Seq data: 1

  • Number of columns with CNV data: 1

Parameters for cross-validation:

  • Number of performed cross-validations: 5

  • Number of folds for cross-validation: 5

Data set - preview:

Statistics of numerical variables

Statistics of nominal variables

  • Variables omitted in a summary: patient_id, karyotype

Survival curve for the entire set

Decision trees

  • The entire data set was used for creating decision trees.

rpart

Nodes description:

  • first line - class

  • second line - number of correct classifications to the number of observations in the node

  • third line - percentage of observations in the node

  • A number at the top of the node (for example [2]) is the id of the path from the root to this node. This number corresponds to the curve with this same id in the plot belowe.

  • Double-click on legend to isolate one trace

ctree

  • A number at the top of the node (for example 2) is the id of the path from the root to this node. This number corresponds to the curve with this same id in the plot belowe.

  • Double-click on legend to isolate one trace

Decision rules

Number of examples in each class:

outcome number of examples
alive 51
dead 58

Decision rules:

Legend:

  • p - number of covered examples with the outcome equal to the rule outcome.
  • n - number of covered examples with the outcome not equal to the rule outcome.
  • Double-click on legend to isolate one trace

Variable importance ranking

The variable importance ranking is calculated on the basis of random forest.

Evaluation of predictors quality

The values in this table are the mean values obtained from repeated k-fold cross-validations.