NEWS

---

multiclassPairs 2021-05-16

minor CRAN fixes

multiclassPairs 2021-01-26

minor changes

• minor change in rule_based_RandomForest print method
• default of k_range in train_one_vs_rest_TSP set to 10:50 instead of 2:50
• default of genes_altogether and genes_one_vs_rest in sort_rules_RF set to 50 instead of 200
• default of rules_altogether and rules_one_vs_rest in train_RF set to 50 instead of 200
• Update the tutorial with time and accuracy comparisons

multiclassPairs 2020-11-19

changes

• train_RF has optimized gene_repetition method

multiclassPairs 2020-11-16

changes

• replace the mode imputation method by kNN method in predict_RF function.
• train_RF now stores the whole binary matrix instead of mode matrix.
• change work-flow figures in the tutorial.
• the predict_RF function can predict matrix with one sample with no error

multiclassPairs 2020-11-02

changes:

• proximity_matrix_RF replaced cocluster_RF function and it can return and plot the proximity matrix

Bug fixes:

• FIXED: plot_binary_RF does not get the predictions and scores when using as_training=TRUE and top_anno="platfrom" or "prediction"

multiclassPairs 2020-10-09

Additions:

• Tutorial is available now.

Minor changes:

• easier access to switchBox disjoint argument in train_one_vs_rest_TSP function.
• Update examples.

Bug fixes:

• plot_binary_TSP when using ExpressionSet as input with no ref or platform.
• passing additional arguments to SB training function by the user.
• printing number of rules in the print function for sorted rules.
• border = NA instead of border = FALSE in plotting functions.
• optimize_RF can handle two classes problems without errors
• num.trees = num.trees missed in ranger for featureNo_altogether slots

multiclassPairs 2020-09-28

Dependencies:

• Dependency issue solved (switchBox and Biobase packages are installed separately).

Minor changes:

• Update examples.

multiclassPairs 2020-09-24

Additions:

• additional function summary_genes_RF to summarize genes to rules stats.
• additional function optimize_RF to help in train_RF parameters optimization.

Changes:

• plot_binary_RF now supports when RF model is trained with probability = FALSE.
• plot_binary_RF extracts prediction labels for training data from the classifier object.
• imputation is implemented in predict_RF function.
• NA is not allowed for class and platforms labels.

Optimizations:

• stats for gene repetition in rules are stored in the sorted rules object to make training process faster.

Minor changes:

• Update examples.
• minor bug fixes.

multiclassPairs 2020-09-08

• first release on CRAN servers