Multi-scale exponential-smoothing forecaster for time series data
System determines, for first value at first time in time series, first estimate based on value and velocity for time series, first lag, and first time. System determines, for first value, second estimate based on value and velocity for time series, second lag, and first time. System determines first weight based on difference between second value, at second time in time series, and first estimate and second weight based on difference between second value and second estimate. System determines, for second value, first forecast based on value and velocity for time series, first lag, and second time. System determines, for second value, second forecast based on value and estimated velocity for time series, second lag, and second time. System determines, for second value, combined forecast based on first forecast weighed by first weight and second forecast weighed by second weight. If combined forecast satisfies threshold, system outputs alert.
1. A system for a multi-scale exponential-smoothing forecaster for time series data, the system comprising:
one or more processors; and
a non-transitory computer readable medium storing a plurality of instructions, which when executed, cause the one or more processors to:
receive time series data indicative of a resource utilization by an application over time, wherein the time series data includes a current value indicative of a current utilization of a resource by the application at a current time;
determine a forecasted future value in the time series data, wherein the determining includes:
determining, for the current value, a first estimate of the current value based on a first prior value and a first velocity indicating a first rate of change associated with the first prior value, wherein the first prior value precedes the current value in the time series data by a first time interval;
determining, for the current value, a second estimate of the current value based on a second prior value and a second velocity indicating a second rate of change associated with the second prior value, wherein the second prior value precedes the current value in the time series data by a second, different time interval;
determining a first weight indicative of an accuracy of the first estimate by comparing the first estimate to the current value;
determining a second weight indicative of an accuracy of the second estimate by comparing the second estimate to the current value; and
forecasting the future value by:
determining first and second estimates of the future value based on the current value and the first prior value; and
combining the first and second estimates of the future value adjusted based on the first and second weights in a manner that implements multi-scale exponential-smoothing; and
provide, via a user interface, an alert associated with the resource utilization based on the determined forecasted future value satisfying a threshold.
2. The system of claim 1 , wherein the plurality of instructions further causes the one or more processors to train a machine learning system to determine at least one forecasted future value for a training set comprising at least some values in the time series data.
3. The system of claim 1 , wherein the resource utilization includes a memory utilization of a cloud-based application.
4. The system of claim 1 , wherein the first estimate of the current value is further based on a first acceleration associated with the first prior value, wherein the first acceleration is a difference between the first velocity and the second velocity.
5. The system of claim 1 , wherein the plurality of instructions further causes the one or more processors to:
determine, for the current value, a third estimate of the current value based on a third prior value and a third velocity indicating a third rate of change associated with the third prior value, wherein the third prior value precedes the current value in the time series data by a third time interval;
determine a third weight indicative of an accuracy of the third estimate by comparing the third estimate to the current value; and
wherein the combining includes combining the first, second, and third estimates of the future value adjusted based on the first, second, and third weights.
6. The system of claim 1 , wherein the resource utilization includes a processor utilization of a cloud-based application.
7. The system of claim 1 , wherein the resource utilization describes a performance metric of a cloud-based application.
8. A non-transitory computer-readable medium have programing instructions for a multi-scale exponential-smoothing forecaster stored therein that are executable by a computing system to:
receive time series data indicative of a resource utilization by an application over time, wherein the time series data includes a current value indicative of a current utilization of a resource by the application at a current time;
determine a forecasted future value in the time series data, wherein the determining includes:
determining, for the current value, a first estimate of the current value based on a first prior value and a first velocity indicating a first rate of change associated with the first prior value, wherein the first prior value precedes the current value in the time series data by a first time interval;
determining, for the current value, a second estimate of the current value based on a second prior value and a second velocity indicating a second rate of change associated with the second prior value, wherein the second prior value precedes the current value in the time series data by a second, different time interval;
determining a first weight indicative of an accuracy of the first estimate by comparing the first estimate to the current value;
determining a second weight indicative of an accuracy of the second estimate by comparing the second estimate to the current value; and
forecasting the future value by:
determining first and second estimates of the future value based on the current value and the first prior value; and
combining the first and second estimates of the future value adjusted based on the first and second weights in a manner that implements multi-scale exponential-smoothing; and
provide an alert via a user interface based on the determined forecasted future value satisfying a threshold.
9. The computer-readable medium of claim 8 , wherein the program instructions are further executable to train a machine learning system to determine at least one forecasted future value for a training set comprising at least some values in the time series data.
10. The computer-readable medium of claim 8 , wherein the resource utilization includes a memory utilization of a cloud-based application.
11. The computer-readable medium of claim 8 , wherein the first estimate of the current value is further based on a first acceleration associated with the first prior value, wherein the first acceleration is a difference between the first velocity and the second velocity.
12. The computer-readable medium of claim 8 , wherein the program instructions are further executable to:
determine, for the current value, a third estimate of the current value based on a third prior value and a third velocity indicating a third rate of change associated with the third prior value, wherein the third prior value precedes the current value in the time series data by a third time interval;
determine a third weight indicative of an accuracy of the third estimate by comparing the third estimate to the current value; and
wherein the combining includes combining the first, second, and third estimates of the future value adjusted based on the first, second, and third weights.
13. The computer-readable medium of claim 8 , wherein the resource utilization includes a processor utilization of a cloud-based application.
14. The computer-readable medium of claim 8 , wherein the resource utilization describes a performance metric of a cloud-based application.
15. A computer-implemented method for a multi-scale exponential-smoothing forecaster for time series data, the method comprising:
receiving time series data indicative of a resource utilization by an application over time, wherein the time series data includes a current value indicative of a current utilization of a resource by the application at a current time; and
providing an alert via a user interface based on a forecasted future value in the time series data satisfying a threshold, wherein determining the forecasted future value includes:
determining, for the current value, a first estimate of the current value based on a first prior value and a first velocity indicating a first rate of change associated with the first prior value, wherein the first prior value precedes the current value in the time series data by a first time interval;
determining, for the current value, a second estimate of the current value based on a second prior value and a second velocity indicating a second rate of change associated with the second prior value, wherein the second prior value precedes the current value in the time series data by a second, different time interval;
determining a first weight indicative of an accuracy of the first estimate by comparing the first estimate to the current value;
determine a second weight indicative of an accuracy of the second estimate by comparing the second estimate to the current value; and
forecasting the future value by:
determining first and second estimates of the future value based on the current value and the first prior value; and
combining the first and second estimates of the future value adjusted based on the first and second weights in a manner that implements multi-scale exponential-smoothing.
16. The computer-implemented method of claim 15 , wherein the computer-implemented method further comprises training a machine learning system to determine at least one forecasted future value for a training set comprising at least some values in the time series data.
17. The computer-implemented method of claim 15 , wherein the resource utilization includes a memory utilization of a cloud-based application.
18. The computer-implemented method of claim 15 , wherein the first estimate of the current value is further based on a first acceleration associated with the first prior value, wherein the first acceleration is a difference between the first velocity and the second velocity.
19. The computer-implemented method of claim 15 , wherein the computer-implemented method further comprises:
determining, for the current value, a third estimate of the current value based on a third prior value and a third velocity indicating a third rate of change associated with the third prior value, wherein the third prior value precedes the current value in the time series data by a third time interval;
determining a third weight indicative of an accuracy of the third estimate by comparing the third estimate to the current value; and
wherein the combining includes combining the first, second, and third estimates of the future value adjusted based on the first, second, and third weights.
20. The computer-implemented method of claim 15 , wherein the resource utilization describes a performance metric of a cloud-based application.