Recommender system

• result of recommender system training: https://yadi.sk/d/PvwkEynop2eC9
• sample: https://github.com/Undin/recommender-system/blob/master/src/main/java/com/ifmo/recommendersystem/Main.java

P.S. there aren't some meta-features for kdd_ipums_la_97-small, kdd_ipums_la_98-small, kdd_ipums_la_99-small, mushroom, pendigits, splice, sylva_agnostic, sylva_prior (knn and neural meta-features).

Files

• algorithms.json - list of used feature subset selection algorithms. Contains short name, full class names and options
• classifiers.json - list of used classifier algorithms. Contains short name, full class names and options
• config.json - config for recommender system builder
• evaluation configs (such as evaluationConfig.json, fastEvaluationConfig.json and etc.) - configs with parameters to use recommender system

How to use

• download results and unzip to root of project
• change DATASETS to array with paths to your datasets
• run Main
• ...
• PROFIT!!! - recommendation will be located in OUTPUT_DIRECTORY

todo

create console util