Accuracy of multivariate approach for time-series based forecasting
In some implementations, a monitoring device may obtain a plurality of time-series data streams respectively associated with a plurality of resources. The monitoring device may generate, using a plurality of machine learning models and based on the plurality of time-series data streams, a plurality of sets of multi-step forecast values, wherein each set of multi-step forecast values is associated with the plurality of resources. The monitoring device may determine, based on the plurality of sets of multi-step forecast values, a set of particular multi-step forecast values associated with the plurality of resources. The monitoring device may cause, based on the set of particular multi-step forecast values, one or more actions to be performed. In some implementations, the monitoring device may determine, based on the plurality of time-series data streams and the plurality of sets of multi-step forecast values, that a correlation exists between a first resource and a second resource.
1 . A device, comprising:
one or more memories; and
one or more processors to:
obtain a plurality of time-series data streams respectively associated with a plurality of resources;
generate, using a first machine learning model and based on the plurality of time-series data streams, a set of first multi-step forecast values associated with the plurality of resources;
generate, using a second machine learning model and based on the plurality of time-series data streams, a set of second multi-step forecast values associated with the plurality of resources;
generate, using a third machine learning model and based on the plurality of time-series data streams, a set of third multi-step forecast values associated with the plurality of resources;
determine, based on the set of first multi-step forecast values, the set of second multi-step forecast values, and the set of third multi-step forecast values, a set of particular multi-step forecast values associated with the plurality of resources; and
cause, based on the set of particular multi-step forecast values, one or more actions to be performed, wherein the one or more actions include at least one of:
adjustment of one or more operation parameters associated with at least one resource of the plurality of resources; or
adjustment of one or more access parameters associated with at least one resource of the plurality of resources.
2 . The device of claim 1 , wherein each of the first machine learning model, the second machine learning model, and the third machine learning model is a multivariate multiple parallel time-series based machine learning model, wherein:
the first machine learning model is trained using a first set of training data;
the second machine learning model is trained using a second set of the training data;
the third machine learning model is trained using a third set of the training data;
at least some of the first set and at least some of the second set do not overlap with each other;
at least some of the second set and at least some of the third set do not overlap with each other; and
at least some of the first set and at least some of the third set do not overlap with each other.
3 . The device of claim 1 , wherein:
the set of first multi-step forecast values associated with the plurality of resources is generated, using the first machine learning model, based on a first set of the plurality of time-series data streams;
the set of second multi-step forecast values associated with the plurality of resources is generated, using the second machine learning model, based on a second set of the plurality of time-series data streams;
the set of third multi-step forecast values associated with the plurality of resources is generated, using the third machine learning model, based on a third set of the plurality of time-series data streams;
at least some of the first set and at least some of the second set do not overlap with each other;
at least some of the second set and at least some of the third set do not overlap with each other; and
at least some of the first set and at least some of the third set do not overlap with each other.
4 . The device of claim 1 , wherein the one or more processors, to determine the set of particular multi-step forecast values, are to:
determine, using a voting technique and based on the set of first multi-step forecast values, the set of second multi-step forecast values, and the set of third multi-step forecast values, the set of particular multi-step forecast values.
5 . The device of claim 1 , wherein the one or more processors, to determine the set of particular multi-step forecast values, are to:
identify a first multi-step forecast value, of the set of first multi-step forecast values, associated with a particular resource of the plurality of resources;
identify a second multi-step forecast value, of the set of second multi-step forecast values, associated with the particular resource;
identify a third multi-step forecast value, of the set of third multi-step forecast values, associated with the particular resource; and
determine, using an averaging technique and based on the first multi-step forecast value, the second multi-step forecast value, and the third multi-step forecast value, a particular multi-step forecast value associated with the particular resource,
wherein the particular multi-step forecast value is part of the set of particular multi-step forecast values.
6 . The device of claim 1 , wherein the one or more processors, to cause the one or more actions to be performed, are to:
provide, to another device, the set of particular multi-step forecast values,
wherein providing the set of particular multi-step forecast values is to permit the other device to display the set of particular multi-step forecast values on a display of the other device.
7 . The device of claim 1 , wherein the one or more processors are further to:
identify a first time-series data stream, of the plurality of time-series data streams, that is associated with a first resource of the plurality of resources;
identify a second time-series data stream, of the plurality of time-series data streams, that is associated with a second resource of the plurality of resources;
identify a first multi-step forecast value, of the set of first multi-step forecast values, associated with the first resource and another first multi-step forecast value associated with the second resource;
identify a second multi-step forecast value, of the set of second multi-step forecast values, associated with the first resource and another second multi-step forecast value associated with the second resource;
identify a third multi-step forecast value, of the set of third multi-step forecast values, associated with the first resource and another third multi-step forecast value associated with the second resource; and
determine, based on the first time-series data stream, the second time-series data stream, the first multi-step forecast value, the second multi-step forecast value, and the third multi-step forecast value, that a correlation exists between the first resource and the second resource.
8 . The device of claim 7 , wherein the one or more processors are further to:
obtain, based on determining that the correlation exists between the first resource and the second resource, another plurality of time-series data streams respectively associated with another plurality of resources,
wherein the other plurality of resources includes the resources of the plurality of resources except the first resource;
generate, using the first machine learning model and based on the other plurality of time-series data streams, a set of other first multi-step forecast values associated with the other plurality of resources;
generate, using the second machine learning model and based on the other plurality of time-series data streams, a set of other second multi-step forecast values associated with the other plurality of resources;
generate, using the third machine learning model and based on the other plurality of time-series data streams, a set of other third multi-step forecast values associated with the plurality of resources;
determine, based on the set of other first multi-step forecast values, the set of other second multi-step forecast values, and the set of other third multi-step forecast values, a set of other particular multi-step forecast values associated with the other plurality of resources;
determine, based on the set of other particular multi-step forecast values, a first other particular multi-step value associated with the first resource; and
cause, based on the set of other particular multi-step forecast values and the first other particular multi-step value, one or more additional actions to be performed.
9 . The device of claim 1 , wherein each of the plurality of time-series data streams are associated with different one of the plurality of resources.
10 . The device of claim 1 , wherein each of the plurality of sets of multi=step forecasts are generated from a different machine learning model.
11 . A non-transitory computer-readable medium storing a set of instructions, the set of instructions comprising:
one or more instructions that, when executed by one or more processors of a device, cause the device to:
obtain a plurality of time-series data streams respectively associated with a plurality of resources;
generate, using three machine learning models and based on the plurality of time-series data streams, a set of first multi-step forecast values associated with the plurality of resources, a set of second multi-step forecast values associated with the plurality of resources, and a set of third multi-step forecast values associated with the plurality of resources;
determine, based on the set of first multi-step forecast values, the set of second multi-step forecast values, and the set of third multi-step forecast values, a set of particular multi-step forecast values associated with the plurality of resources; and
cause, based on the set of particular multi-step forecast values, one or more actions to be performed, wherein the one or more actions include at least one of:
adjustment of one or more operation parameters associated with at least one resource of the plurality of resources; or
adjustment of one or more access parameters associated with at least one resource of the plurality of resources.
12 . The non-transitory computer-readable medium of claim 11 , wherein the three machine learning models are different from each other.
13 . The non-transitory computer-readable medium of claim 11 , wherein the one or more instructions, that cause the device to determine the set of particular multi-step forecast values, cause the device to:
select the set of particular multi-step forecast values from the set of first multi-step forecast values, the set of second multi-step forecast values, and the set of third multi-step forecast values,
wherein a particular multi-step forecast value, of the set of particular multi-step forecast values, that is associated with a resource, of the plurality of resources, is selected from one of:
a first multi-step forecast value, of the set of first multi-step forecast values, that is associated with the resource;
a second multi-step forecast value, of the set of second multi-step forecast values, that is associated with the resource; and
a third multi-step forecast value, of the set of third multi-step forecast values, that is associated with the resource.
14 . The non-transitory computer-readable medium of claim 11 , wherein the one or more instructions, that cause the device to cause the one or more actions to be performed, cause the device to:
generate a message that includes information associated with the set of particular multi-step forecast values; and
provide, to another device, the message,
wherein providing the message is to permit a user of the other device to be informed of the information associated with the set of particular multi-step forecast values.
15 . The non-transitory computer-readable medium of claim 11 , wherein the one or more instructions, when executed by the one or more processors, further cause the device to:
determine, based on the plurality of time-series data streams, the set of first multi-step forecast values, the set of second multi-step forecast values, and the set of third multi-step forecast values, that a correlation exists between a first resource and a second resource of the plurality of resources.
16 . The non-transitory computer-readable medium of claim 15 , wherein the one or more instructions, when executed by the one or more processors, further cause the device to:
obtain, based on determining that the correlation exists between the first resource and the second resource, another plurality of time-series data streams respectively associated with another plurality of resources,
wherein the other plurality of resources includes the resources of the plurality of resources except the first resource;
generate, using the three machine learning models and based on the other plurality of time-series data streams, a set of other first multi-step forecast values associated with the other plurality of resources, a set of other second multi-step forecast values associated with the other plurality of resources, and a set of other third multi-step forecast values associated with the plurality of resources;
determine, based on the set of other first multi-step forecast values, the set of other second multi-step forecast values, and the set of other third multi-step forecast values, a set of other particular multi-step forecast values associated with the other plurality of resources;
determine, based on the set of other particular multi-step forecast values, a first other particular multi-step value associated with the first resource; and
cause, based on the set of other particular multi-step forecast values and the first other particular multi-step value, one or more additional actions to be performed.
17 . The non-transitory computer-readable medium of claim 16 , wherein the one or more instructions, that cause the device to cause the one or more additional actions to be performed, cause the device to:
provide, to another device, the set of other particular multi-step forecast values and the first other particular multi-step value,
wherein providing the set of other particular multi-step forecast values and the first other particular multi-step value is to permit the other device to display the set of other particular multi-step forecast values and the first other particular multi-step value on a display of the other device.
18 . A method, comprising:
obtaining a plurality of time-series data streams respectively associated with a plurality of resources;
generating, using a plurality of machine learning models and based on the plurality of time-series data streams, a plurality of sets of multi-step forecast values,
wherein each set of multi-step forecast values is associated with the plurality of resources;
determining, based on the plurality of sets of multi-step forecast values, a set of particular multi-step forecast values associated with the plurality of resources;
determining, based on the plurality of time-series data streams and the plurality of sets of multi-step forecast values, that a correlation exists between a first resource and a second resource of the plurality of resources; and
causing, based on the set of particular multi-step forecast values, one or more actions to be performed.
19 . The method of claim 18 , further comprising:
obtaining, based on determining that the correlation exists between the first resource and the second resource, another plurality of time-series data streams respectively associated with another plurality of resources,
wherein the other plurality of resources includes the resources of the plurality of resources except the first resource;
generating, using the plurality of machine learning models and based on the other plurality of time-series data streams, a plurality of sets of other multi-step forecast values,
wherein each set of other multi-step forecast values is associated with the other plurality of resources;
determining, based on the plurality of sets of other multi-step forecast values, a set of other particular multi-step forecast values associated with the other plurality of resources;
determining, based on the set of other particular multi-step forecast values, a first other particular multi-step value associated with the first resource; and
causing, based on the set of other particular multi-step forecast values and the first other particular multi-step value, one or more additional actions to be performed.
20 . The method of claim 19 , further comprising:
removing, based on determining that the correlation exists between the first resource and the second resource and prior to generating the plurality of other sets of multi-step forecast values, at least one machine learning model from the plurality of machine learning models,
wherein the plurality of other sets of multi-step forecast values are generated using other machine learning models that remain in the plurality of machine learning models.