Analytics-based anomaly detection

1. An apparatus, comprising:

a processor; and

a non-volatile memory that stores code executable by the processor to:

receive measurement data for a component of a device in real-time, the measurement data comprising measurements for at least one condition of the component that is monitored over a defined time interval;

determine, dynamically, a compression rate of the received real-time measurement data by applying a compression algorithm to the measurement data, the compression rate indicating a compressed size of the measurement data as a result of applying the compression algorithm relative to an original size of the measurement data; and

generate an alert indicating a possible anomaly in an operation of the component represented by the measurement data in response to the determined compression rate satisfying a threshold compression rate for the at least one condition of the component, wherein satisfying the threshold compression rate indicates an abnormal operating state of the component.

2. The apparatus of claim 1 , wherein the compression algorithm that is used to determine the compression rate for the measurement data is configured to create a compressed data set by identifying and storing values that fall outside of a range that is dynamically calculated for each value in the measurement data based on a height of the measurement data.

3. The apparatus of claim 1 , wherein the compression algorithm determines the compression rate by:

determining a minimum value and a maximum value of the measurement data;

calculating a height of the measurement data by determining a difference between the maximum value and the minimum value;

determining a delta comprising a difference in level that determines whether a value is removed from the measurement data, the delta determined using a logarithmic function of the calculated height;

creating the compressed data set by, for every value in the measurement data, removing a current value from the measurement data in response to the current value falling within the range defined as (previous value−delta)<the current value <(previous value+delta), wherein the previous value is a most recent value placed in the compressed data set; and

determining the compression rate by calculating (100−(length of the compressed data set)*100/length of measurement data).

4. The apparatus of claim 1 , wherein the code is executable by the processor to:

receive new measurement data for the component on a continuous basis;

determine a new compression rate, using the compression algorithm, based on the received new measurement data; and

generate an alert in response to the determined new compression rate satisfying the threshold compression rate for the at least one condition of the component.

5. The apparatus of claim 1 , wherein the code is executable by the processor to:

check an operating state of at least one other component of the device that is related to the component for a same time interval in response to the determined compression rate satisfying the threshold compression rate; and

generate the alert in response to the operating state being normal for the at least one other component of the device for the same time interval.

6. The apparatus of claim 1 , wherein the code is executable by the processor to:

check for system alerts associated with the component that are triggered during a same time interval in response to the determined compression rate satisfying the threshold compression rate; and

generate the alert in response to other system alerts associated with the component being triggered for the same time interval.

7. The apparatus of claim 1 , wherein the code is executable by the processor to use machine learning to analyze the measurement data and the determined compression rate to generate at least one prediction associated with the operating state of the component, the prediction associated with an operating state of the component and provided as part of the generated alert.

8. The apparatus of claim 7 , wherein the machine learning is trained using training data that comprises measurement and other operating data from a plurality of different devices that have components that are comparable to the components of the device.

9. The apparatus of claim 7 , wherein the code is executable by the processor to predict, using the machine learning, an estimate of when the component that generated the alert will fail, the determined estimate provided as part of the generated alert.

10. The apparatus of claim 7 , wherein the code is executable by the processor to predict, using the machine learning and based on the measurement data and the determined compression rate, a cost of operation for the device, the cost of operation comprising a per-user cost of the device in a hardware-as-a-service environment.

11. The apparatus of claim 1 , wherein the code is executable by the processor to determine a severity level for the generated alert based on the determined compression rate relative to one of a plurality of different compression rate thresholds, wherein different severity levels are each associated with one of the plurality of different compression rate thresholds.

12. The apparatus of claim 11 , wherein the code is executable by the processor to determine one or more recommendations associated with the component based on the severity level for the generated alert.

13. The apparatus of claim 1 , wherein the code is executable by the processor to determine the threshold compression rate based on at least one of user input, a set of baseline measurement data that is received when the component is installed, manufacturer specifications for the component, and measurement data for related components.

14. The apparatus of claim 1 , wherein the code is executable by the processor to include details about at least one specification of the component as part of the generated alert and provide at least one recommendation for correcting the possible anomaly in the component based on the details about the at least one specification.

15. A method, comprising:

receiving, by a processor, measurement data for a component of a device in real-time, the measurement data comprising measurements for at least one condition of the component that is monitored over a defined time interval;

determining, dynamically, a compression rate of the received real-time measurement data by applying a compression algorithm to the measurement data, the compression rate indicating a compressed size of the measurement data as a result of applying the compression algorithm relative to an original size of the measurement data; and

generating an alert indicating a possible anomaly in an operation of the component represented by the measurement data in response to the determined compression rate satisfying a threshold compression rate for the at least one condition of the component, wherein satisfying the threshold compression rate indicates an abnormal operating state of the component.

16. The method of claim 15 , wherein the compression algorithm that is used to determine the compression rate for the measurement data is configured to create a compressed data set by identifying and storing values that fall outside of a range that is dynamically calculated for each value in the measurement data based on a height of the measurement data.

17. The method of claim 15 , further comprising:

receiving new measurement data for the component on a continuous basis;

determining a new compression rate, using the compression algorithm, based on the received new measurement data; and

generating an alert in response to the determined new compression rate satisfying the threshold compression rate for the at least one condition of the component.

18. The method of claim 15 , further comprising:

checking for system alerts associated with the component that are triggered during a same time interval in response to the determined compression rate satisfying the threshold compression rate; and

generating the alert in response to other system alerts associated with the component being triggered for the same time interval.

19. The method of claim 15 , further comprising using machine learning to analyze the measurement data and the determined compression rate to generate at least one prediction associated with the operating state of the component, the prediction associated with an operating state of the component and provided as part of the generated alert.

20. A program product comprising a non-volatile computer readable storage medium and program code, the program code being configured to be executable by a processor to perform operations comprising:

receiving measurement data for a component of a device in real-time, the measurement data comprising measurements for at least one condition of the component that is monitored over a defined time interval;

determining, dynamically, a compression rate of the received real-time measurement data by applying a compression algorithm to the measurement data, the compression rate indicating a compressed size of the measurement data as a result of applying the compression algorithm relative to an original size of the measurement data; and

generating an alert indicating a possible anomaly in an operation of the component represented by the measurement data in response to the determined compression rate satisfying a threshold compression rate for the at least one condition of the component, wherein satisfying the threshold compression rate indicates an abnormal operating state of the component.