Supporting data for "Automatic and Efficient Privacy Preserving and Fault Detection Techniques for Big-data Systems"

This dataset consists of all the experiment data of my three works: UPA, GUPA and Themis. Specifically, this dataset shows the accuracy (in inferring sensitivity and computing the final output of a big-data query), efficiency (execution time) and scalability (execution time variation due to differences in dataset sizes, sample sizes etc) of UPA and GUP. This dataset also shows the fault detection capability (the correlation between the number of faults detected from a DLS and the DLS’s error rate), number of faults detected and retraining accuracy of Themis.