PMC:4289553 / 22314-23036 JSONTXT

{"target":"http://pubannotation.org/docs/sourcedb/PMC/sourceid/4289553","sourcedb":"PMC","sourceid":"4289553","source_url":"https://www.ncbi.nlm.nih.gov/pmc/4289553","text":"Another study compared the performance of 6 data mining tools at various sample sizes for 2 test databases (test database I with 50,000 records and test data base II with 1,500,000 records), using accuracy as the performance measure. For test database I, for all tools, a stable level of accuracy was reached at 16,000 records, and for test database II, for all tools, a stable level was reached at 8000 records [10]. The results of our study are in line with these studies. Although we used mean validated AUC-values instead of accuracy to measure the performance of the models, we also found that the more complex modelling techniques required large numbers of events per variable to generate models with optimism \u003c0.01.","tracks":[]}