Resource data modeling, forecasting and simulation

Patent metadata

Field	Value
Publication number	US-2022027744-A1
Application number	US-202016936184-A
Country	US
Kind code	A1
Filing date	Jul 22, 2020
Priority date	Jul 22, 2020
Publication date	Jan 27, 2022
Grant date	—

Abstract

A resource data modeling, forecasting, and simulation system analyzes data pertaining to the data processing tasks and the resources assigned to the data processing tasks to generate short-term forecasts and long-term forecasts of task volumes. The forecasted task volumes are further optimized based on different factors to determine the resources required to handle the forecasted task volume. Various simulations of hypothetical what-if scenarios are also generated based on the forecasts and the resource requirements. The resource data modeling, forecasting and simulation system is based on multi-algorithmic ensemble models for forecasting, automated model selection and the unique simulation methodology based on multiple parameters.

First claim

What is claimed is:

1 . A resource data modeling and simulation system, comprising:
at least one processor; a non-transitory processor-readable medium storing machine-readable instructions that cause the processor to: receive historical time series data regarding data processing tasks and resources that are employed for handling the data processing tasks; receive input regarding one of a short-term forecast and a long-term forecast to be generated, the short-term forecast and the long-term forecast pertaining to a number of the data processing tasks expected to be received in a corresponding one of a shorter time period and a longer time period; if the input specifies the short-term forecast, then:
obtain short-term aggregated data by aggregating the time series data for the shorter time period;
generate updated short-term aggregated data including short-term aggregated data and aggregated data obtained from a time interval that elapsed since the short-term aggregated data was generated;
if an accuracy of an output of a prior short-term forecasting model falls within a predetermined accuracy threshold,
obtain the short-term forecast by employing the prior short-term forecasting model for analyzing the updated short-term aggregated data;

if an accuracy of the output of the prior short-term forecasting model falls outside the predetermined threshold,
obtain the short-term forecast by employing a new short-term forecasting model for analyzing the updated short-term aggregated data;

and if the input requires the long-term forecast, then:
obtain long-term aggregated data by aggregating the time series data for the longer time period;
generate updated long-term aggregated data including the long-term aggregated data and aggregated data obtained from a time interval since the long-term aggregated data was generated and the specified time;
select a long-term forecasting model from a plurality of long-term forecasting models based on an accuracy of an output of the selected long-term forecasting model falling within the predetermined threshold;
obtain the long-term forecast by employing the selected long-term forecasting model for analyzing the updated long-term aggregated data;
calculate a number of resources required to handle the data processing tasks expected to be received in the corresponding one of the shorter time period and the longer time period; and generate simulations for one or more parameters associated with the data processing tasks in response to simulation user input based at least on optimized values of one or more of the short-term forecast and the long-term forecast.

2 . The resource data modeling and simulation system of claim 1 , wherein the historical time series data includes univariate time series data of requests for the data processing tasks, average handling time (AHT) for each of the data processing tasks, shift duration of the resources, productive hours including a time period for which the resources handled the data processing tasks and shrinkage values associated with the shifts.

3 . The resource data modeling and simulation system of claim 1 , wherein the processor is to further:
preprocess the historical time series data for one or more of missing values and outliers; and obtain a hypertext markup language (HTML) file including the historical time series data.

4 . The resource data modeling and simulation system of claim 1 , wherein to obtain the short-term forecast by employing the prior short-term forecasting model, the processor is to further:
calculate a mean absolute percentage error (MAPE) for the prior short-term forecasting model; and employ the prior short-term forecasting model for generating the short-term forecast if the MAPE of the prior forecasting model is less than a predetermined percentage or if the MAPE of the prior short-term forecasting model is less than a predetermined point variation from a prior MAPE of the prior forecasting model.

5 . The resource data modeling and simulation system of claim 1 , wherein to obtain the short-term forecast by employing the new short-term forecasting model the processor is to further:
split the short-term aggregated data into training data and testing data; train a plurality of forecasting models on the training data; test the plurality of forecasting models on the testing data; compare accuracies of the plurality of forecasting models; and select as the new short-term forecasting model, one of the plurality of forecasting models with highest accuracy.

6 . The resource data modeling and simulation system of claim 5 , wherein to select the new short-term forecasting model the processor is to further:
calculate a mean absolute percentage error (MAPE) for each of the plurality of forecasting models; and select one of the plurality of forecasting models with a minimum MAPE value as the new short-term forecasting model.

7 . The resource data modeling and simulation system of claim 1 , wherein to obtain the optimized values the processor is to further:
generate the short-term aggregated data by aggregating the historical time series data at a daily level; obtain the short-term forecast at the daily level using the short-term aggregated data aggregated at the daily level; and calculate data proportions for hourly and half-hourly periods from the short-term forecast at the daily level.

8 . The resource data modeling and simulation system of claim 7 , wherein to obtain the optimized values of the forecasts the processor is to further:
split the short-term forecast at the daily level at one of the hourly and the half-hourly periods based on the data proportions; and compute a number of the resources based on a forecasted task volume obtained from splitting the short-term forecast at the daily level.

9 . The resource data modeling and simulation system of claim 8 , wherein to obtain the optimized value of the short-term forecast the processor is to further:
create a decision matrix using the number of resources computed based on the forecasted task volume in conjunction with shift timings of the resources, wherein the shift timings are obtained from a shift information source file; calculate a number of resources required hourly for completion of the data processing tasks by using linear programming with shifts specified in the shift information source file; and further, optimize the number of resources required hourly based on a shrinkage factor.

10 . The resource data modeling and simulation system of claim 1 , wherein to obtain the optimized values of the forecasts for long-term the processor is to further:
generate the long-term aggregated data by aggregating the historical time series data for a plurality of weeks; and calculate the optimized value of the long-term forecast using a formula:

the
⁢

⁢
optimized
⁢

⁢
value
⁢

⁢
of
⁢

⁢
the
⁢

⁢
long
⁢

-

⁢
term
⁢

Assignees

• Accenture Global Solutions Ltd

Inventors

• Krishnan Mythili
• Ambhore Dnyaneshwar
• Bhowmick Kunal
• Sundar Lalitha
• Nair Vinita
• Purushothaman Arun
• Janardhanan Anand
• Rao Bhavana

Classifications

• G06Q10/06315

  Needs-based resource requirements planning or analysis · CPC title

• G06N20/00 Primary

  Machine learning · CPC title

• G06N5/02 Primary

  Knowledge representation; Symbolic representation · CPC title

• G06F9/455

  Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines · CPC title