Alignment and deduplication of time-series datasets

1. A non-transitory computer-readable storage medium comprising instructions executable by a processor to:

temporally align a group of time-series datasets received from a system;

determine a correlation between a first time-series dataset and a second time-series dataset from the temporally aligned group of time-series datasets;

determine whether an absolute value of the correlation exceeds a correlation threshold;

in response to a determination that the absolute value of the correlation exceeds the correlation threshold, remove the second time-series dataset from the temporally aligned group of time-series datasets to obtain a reduced temporally aligned group of time-series datasets; and

identify an anomaly that has occurred in the system based on the reduced temporally aligned group of time-series datasets.

2. The non-transitory computer-readable storage medium of claim 1 further comprising instructions executable by the processor to, for each time-series dataset of the group of time-series datasets, determine whether a breach has occurred in the time-series dataset.

3. The non-transitory computer-readable storage medium of claim 2 wherein the instructions to temporally align comprise instructions executable by the processor to temporally align the group of time-series datasets in response to a determination that the breach has occurred in each time-series dataset of the group of time-series datasets.

4. The non-transitory computer readable storage medium of claim 1 further comprising instructions executable by the processor to determine the correlation using a Pearson product-moment correlation coefficient.

5. The non-transitory computer-readable storage medium of claim 2 further comprising instructions executable by the processor to:

determine, for the reduced temporally aligned group of time-series datasets, a proportion of the time-series datasets in which the breach has occurred relative to the time-series datasets in which the breach has not occurred; and

determine whether the anomaly has occurred in the system based on the proportion.

6. The non-transitory computer-readable storage medium of claim 5 wherein the instructions to determine whether the anomaly has occurred based on the proportion comprise instructions executable by the processor to:

in response to the proportion exceeding a proportion threshold, determine that the anomaly has occurred in the system; and

in response to the proportion not exceeding the proportion threshold, determine that the anomaly has not occurred in the system.

7. The non-transitory computer-readable storage medium of claim 1 further comprising instructions executable by the processor to:

determine a correlation between each pair of time-series datasets in the temporally aligned group of time-series datasets;

for each pair:

determine whether an absolute value of the correlation between the pair exceeds the correlation threshold; and

in response to a determination that the absolute value of the correlation between the pair exceeds the correlation threshold, remove one time-series dataset of the pair from the temporally aligned group of time-series datasets.

8. The non-transitory computer-readable storage medium of claim 1 further comprising instructions executable by the processor to receive the group of time-series datasets from data sources in the system,

the system being a computer network.

9. The non-transitory computer-readable storage medium of claim 1 wherein the instructions to temporally align comprise instructions executable by the processor to:

determine a first degree of smoothness of the first time-series dataset and a second degree of smoothness of the second time-series dataset;

compare each of the first and second degrees of smoothness with a smoothness threshold;

select one of a plurality of temporal alignment processes based on the comparisons of the first and second degrees of smoothness with the smoothness threshold; and

process at least one of the first and second time-series datasets according to the selected one of the plurality of temporal alignment processes such that a first set of data values in the first time-series dataset become temporally aligned with a second set of data values in the second time-series dataset.

10. The non-transitory computer-readable storage medium of claim 9 wherein:

the plurality of temporal alignment processes comprises:

a first process corresponding to the first and second degrees of smoothness exceeding the smoothness threshold,

a second process corresponding to the first degree of smoothness exceeding the smoothness threshold and the second degree of smoothness not exceeding the smoothness threshold, and

a third process corresponding to the first and second degrees of smoothness not exceeding the smoothness threshold, and

the non-transitory computer-readable storage medium comprises instructions executable by the processor to select among the first, second, or third processes based on a correspondence between the selected process and the first and second degrees of smoothness.

11. The non-transitory computer-readable storage medium of claim 9 , wherein the instructions to select comprise instructions executable by the processor to, in response to the first degree of smoothness exceeding the smoothness threshold, select the one of the plurality of temporal alignment processes,

wherein the selected one of the plurality of temporal alignment processes comprises generating an interpolated data value associated with a first time stamp of the first time-series dataset based on a second time stamp of the second time-series dataset, the first and second time stamps having a same time value.

12. The non-transitory computer-readable storage medium of claim 9 , wherein the instructions to select comprise instructions executable by the processor to, in response to the first and second degrees of smoothness exceeding the smoothness threshold, select the one of the plurality of temporal alignment processes,

wherein the selected one of the plurality of temporal alignment processes comprises:

generating a first interpolated data value associated with a first time stamp of the first time-series dataset based on a second time stamp of the second time-series dataset, the first and second time stamps having a first same time value; and

generating a second interpolated data value associated with a third time stamp of the second time-series dataset based on a fourth time stamp of the first time-series dataset, the third and fourth time stamps having a second same time value.

13. The non-transitory computer-readable storage medium of claim 9 wherein the instructions to select comprise instructions executable by the processor to, in response to the first and second degrees of smoothness not exceeding the smoothness threshold, select the one of the plurality of temporal alignment processes,

wherein the selected one of the plurality of temporal alignment processes comprises:

determining a temporal tolerance based on time spacings between consecutive time stamps in the first and second time-series datasets;

identifying pairs of time stamps in the first and second time-series datasets that are within the determined temporal tolerance of each another; and

including paired data values in the first and second time-series datasets associated with the identified pairs of time stamps.

14. A system comprising:

a processor; and

a memory comprising instructions executable by the processor to:

temporally align a group of time-series datasets received from a data source;

determine a correlation between a first time-series dataset and a second time-series dataset from the temporally aligned group of time-series datasets;

determine whether an absolute value of the correlation exceeds a correlation threshold;

in response to a determination that the absolute value of the correlation exceeds the correlation threshold, remove the second time-series dataset from the temporally aligned group of time-series datasets to obtain a reduced temporally aligned group of time-series datasets; and

determine that an anomaly has occurred in the data source based on the reduced temporally aligned group of time-series datasets.

15. A method comprising:

by a processor:

temporally aligning pairwise combinations of time-series datasets of a group of time-series datasets representing at least one metric of a monitored system;

determining a correlation between a first time-series dataset and a second time-series dataset from the temporally aligned group of time-series datasets;

determining whether an absolute value of the correlation exceeds a correlation threshold;

in response to a determination that the absolute value of the correlation exceeds the correlation threshold, removing the second time-series dataset from the temporally aligned group of time-series datasets to obtain a reduced temporally aligned group of time-series datasets; and

identifying an anomaly associated with the at least one metric of the monitored system based on the reduced temporally aligned group of time-series datasets.

16. The method of claim 15 , comprising:

determining, for the reduced temporally aligned group of time-series datasets, a proportion of the time-series datasets in which a breach has occurred relative to the time-series datasets in which the breach has not occurred; and

determining whether the anomaly has occurred in the system based on the proportion.

17. The method of claim 16 , wherein determining whether the anomaly has occurred comprises:

in response to the proportion exceeding a proportion threshold, determining that the anomaly has occurred in the system.

18. The method of claim 16 , wherein determining whether the anomaly has occurred comprises:

in response to the proportion not exceeding a proportion threshold, determining that the anomaly has not occurred in the system.

19. The system of claim 14 , the instructions executable by the processor to:

determine, for the reduced temporally aligned group of time-series datasets, a proportion of the time-series datasets in which a breach has occurred relative to the time-series datasets in which the breach has not occurred; and

determine whether the anomaly has occurred in the system based on the proportion.

20. The system of claim 19 , the instructions executable by the processor to:

in response to the proportion exceeding a proportion threshold, determine that the anomaly has occurred in the system; and

in response to the proportion not exceeding the proportion threshold, determining that the anomaly has not occurred in the system.