System and method for predicting service metrics using historical data
A method for allocating resources for a plurality of time intervals, including: receiving a forecasted workload and at least one required service metric value; applying a search algorithm to identify an initial allocation assignment; inputting the assignment to a machine learning algorithm, the machine learning algorithm trained on historic data of past intervals; predicting an expected service metric value provided by the initial allocation assignment; adjusting the initial allocation assignment based on a difference between the expected service metric value and the corresponding required service metric value; iteratively repeating the applying, inputting, predicting, and adjusting operations until one of: the expected service metric value predicted for an adjusted allocation assignment is within a predetermined distance of the corresponding at least one required service metric value for the interval; or a predetermined time has elapsed.
1. A method for allocating resources for a plurality of time intervals, the method comprising:
receiving, by an input layer of a machine learning model, a forecasted workload and at least one required service metric value for each of the plurality of time intervals, wherein the forecasted workload is included in a multivariate time series matrix, wherein the time series matrix has a number of cells corresponding to a product of: a number of the plurality of time intervals, and one or more features;
for each interval:
applying a search algorithm to identify an initial allocation assignment, wherein the initial allocation assignment comprises a count vector, and wherein the initial allocation assignment represents how many resources are suggested for each of a plurality of resource skills;
inputting the initial allocation assignment to a machine learning algorithm, wherein the machine learning algorithm has been previously trained on historic data of a plurality of past intervals, and wherein the machine learning algorithm comprises propagating the received workload through a sequence of dense layers in the machine learning model, wherein each of the dense layers is followed by a sigmoid activation;
predicting, for each at least one required service metric, by the machine learning algorithm, an expected service metric value provided by the initial allocation assignment based on at least one element from the multivariate time series matrix;
adjusting, by the search algorithm, the initial allocation assignment based on a difference between the expected service metric value and the corresponding at least one required service metric value, the adjusting using one or more scalar correction ratios for each of the resource skills;
iteratively repeating the applying, inputting, predicting, and adjusting operations until one of:
the expected service metric value predicted for an adjusted allocation assignment is within a predetermined distance of the corresponding at least one required service metric value for the interval; or
a predetermined time has elapsed; and
automatically producing a schedule based on the adjusted allocation assignment.
2. The method of claim 1 , comprising generating, from the iteratively adjusted allocation assignments, an allocation assignment plan for the plurality of time intervals.
3. The method of claim 1 , wherein resources are classified by at least one skill.
4. The method of claim 3 , wherein the forecasted workload comprises a workload broken down into one or more required resource skills for each of the plurality of time intervals.
5. The method of claim 1 , wherein the forecasted workload comprises a volume of incoming communications.
6. The method of claim 5 , wherein at least one incoming communication is chosen from a list comprising: short message service (SMS), web chat, and email.
7. The method of claim 1 , wherein at least one required service metric is chosen from a list comprising: average speed of answer, service level agreement, abandoned percentage, chat latency, and maximum occupancy.
8. The method of claim 1 , wherein the adjusting is based on a correction ratio determined by the equation:
correction
ratio
=
(
1
+
Expected
Service
Metric
Value
)
(
1
+
Required
Service
Metric
Value
)
.
9. The method of claim 1 , wherein the machine learning algorithm comprises at least one of: a regression algorithm, a deep learning algorithm; a neural network; a fully connected neural network; or a convolutional neural network.
10. A system for allocating resources for a plurality of given time intervals, the system comprising:
a memory; and
a processor configured to:
receive, by an input layer of a machine learning model, a forecasted workload and at least one required service metric value for each of the plurality of time intervals, wherein the forecasted workload is included in a multivariate time series matrix, wherein the time series matrix has a number of cells corresponding to a product of: a number of the plurality of time intervals, and one or more features;
for each interval:
apply a search algorithm to identify an initial allocation assignment, wherein the initial allocation assignment comprises a count vector, and wherein the initial allocation assignment represents how many resources are suggested for each of a plurality of resource skills;
apply a machine learning algorithm to the initial allocation assignment to predict, for each at least one required service metric, an expected service metric value provided by the initial allocation assignment based on at least one element from the multivariate time series matrix, wherein the machine learning algorithm comprises propagating the received workload through a sequence of dense layers in the machine learning model, wherein each of the dense layers is followed by a sigmoid activation;
adjust the initial allocation assignment based on a difference between the expected service metric value and the corresponding at least one required service metric value, the adjusting using one or more scalar correction ratios for each of the resource skills;
iteratively repeat the applying, predicting, and adjusting operations until either:
the expected service metric value predicted for an adjusted allocation assignment is within a predetermined distance of the corresponding at least one required service metric value for the interval; or
a predetermined time has elapsed; and
automatically producing a schedule based on the adjusted allocation assignment.
11. The system of claim 10 , wherein the processor is configured to generate, from the iteratively adjusted allocation assignments, an allocation assignment plan for the plurality of time intervals.
12. The system of claim 10 , wherein the machine learning algorithm has been previously trained on historic data of a plurality of past intervals.
13. The system of claim 10 , wherein the processor classifies resources by at least one skill.
14. The system of claim 13 , wherein the received forecasted workload comprises a workload broken down into at least two required resource skills for each of the plurality of time intervals.
15. The system of claim 10 , wherein the processor is configured to adjust the initial allocation assignment based on a correction ratio determined by the equation:
correction
ratio
=
(
1
+
Expected
Service
Metric
Value
)
(
1
+
Required
Service
Metric
Value
)
.