System and method of cognitive risk management

1. A risk management visualization system, comprising:

a system architecture having a UI layer, a backend layer, a solvers layer and at least one server;

the at least one server, each comprising a processor and memory, is configured to:

request, via a first asynchronous application programming interface call to one or more solvers of the solvers layer, to open a supply chain model of a supply chain network;

check, via a second application programming interface call from the backend layer at regular intervals to the one or more solvers of the solvers layer, whether the one or more solvers are busy processing any request or are available to handle the request;

model, using the backend layer, the supply chain network as a supply chain planning problem, wherein one or more key process indicators (KPIs) of the supply chain planning problem are based, at least in part, on one or more input variables, wherein the backend layer uses one or more data update and retrieval Application Programming Interfaces (APIs) to trigger the one or more solvers to fetch data for the modelling;

model, via an application programming interface call from the backend layer to the solvers layer, an impact on the one or more KPIs from each of the one or more input variables at a selected confidence interval using a Bayesian optimization process; and

display, using the UI layer, by one or more call back scripts that transmit requests to the solvers layer and by one or more python script modules, a visualization of a risk profile for the one or more KPIs, the visualization indicating a probability that an actual KPI value differs from a predicted KPI value.

2. The risk management visualization system of claim 1 , wherein the computer is further configured to:

receive a risk range selection, the risk range selection indicating a level of acceptable risk associated with the predicted KPI value;

calculate an optimal value for the one or more input variables and a range of feasible values for the one or more input variables, wherein the feasible values of the one or more input variables generate the predicted KPI value with a risk within the selected risk range; and

display a visualization of the risk profile with the selected risk range.

3. The risk management visualization system of claim 2 , wherein the computer calculates the range of feasible values using a KPI lower bound and KPI upper bound of the selected risk range based on a base KPI value by:

calculating a standard deviation and a mean of the predicted KPI values; and

when the standard deviation is non-zero, setting the KPI lower bound and the KPI upper bound equal to a Gaussian interval for the standard deviation, the mean, and the selected confidence interval.

4. The risk management visualization system of claim 1 , wherein the computer is further configured to:

identify sample points for building a surrogate model using an acquisition function; and

determine the sample points from the acquisition function by calculating input values having high uncertainties and input values having high derivative values.

5. The risk management visualization system of claim 1 , wherein the computer is further configured to:

build a surrogate model by approximating a relationship between each of the one or more input variables and a resulting KPI value, wherein the surrogate model is constructed using the Bayesian optimization process.

6. The risk management visualization system of claim 1 , wherein the visualization of the risk profile further comprises a visualization of a likelihood at a current input value, wherein the current input value is calculated as a difference of a cumulative distribution function from a Gaussian interval for a mean and standard deviation at an optimal input value and at a set sequential percentage value from the optimal input value.

7. The risk management visualization system of claim 1 , wherein the one or more KPIs comprise one or more of:

a network profit;

a transportation cost; and

an overhead cost.

8. A computer-implemented method of risk management visualization, comprising:

requesting, via a first asynchronous application programming interface call to one or more solvers of a solvers layer, to open a supply chain model of a supply chain network;

checking, via a second application programming interface call from the backend layer at regular intervals to the one or more solvers of the solvers layer, whether the one or more solvers are busy processing any request or are available to handle the request;

modeling, using a backend layer, by at least one server, each comprising a processor and memory, the supply chain network as a supply chain planning problem, one or more key process indicators (KPIs) of the supply chain planning problem are based, at least in part, on the one or more input variables, wherein the backend layer uses one or more data update and retrieval Application Programming Interfaces (APIs) to trigger the one or more solvers to fetch data for the modelling;

modeling, via an application programming interface call from the backend layer to the solvers layer, by the at least one server, an impact on the one or more KPIs from each of the one or more input variables at a selected confidence interval using a Bayesian optimization process; and

displaying, using a UI layer, by one or more call back scripts that transmit requests to the solvers layer and by one or more python script modules, by the at least one server, a visualization of a risk profile for the one or more KPIs, the visualization indicating a probability that an actual KPI value differs from a predicted KPI value.

9. The computer-implemented method of claim 8 , further comprising:

receiving, by the computer, a risk range selection, the risk range selection indicating a level of acceptable risk associated with the predicted KPI value;

calculating, by the computer, an optimal value for the one or more input variables and a range of feasible values for the one or more input variables, wherein the feasible values of the one or more input variables generate the predicted KPI value with a risk within the selected risk range; and

displaying, by the computer, a visualization of the risk profile with the selected risk range.

10. The computer-implemented method of claim 8 , further comprising:

identifying, by the computer, sample points for building a surrogate model using an acquisition function; and

determining, by the computer, the sample points from the acquisition function by calculating input values having high uncertainties and input values having high derivative values.

11. The computer-implemented method of claim 8 , further comprising:

building, by the computer, a surrogate model by approximating a relationship between each of the one or more input variables and a resulting KPI value, wherein the surrogate model is constructed using the Bayesian optimization process.

12. The computer-implemented method of claim 8 , wherein the visualization of the risk profile further comprises a visualization of a likelihood at a current input value, wherein the current input value is calculated as a difference of a cumulative distribution function from a Gaussian interval for a mean and standard deviation at an optimal input value and at a set sequential percentage value from the optimal input value.

13. The computer-implemented method of claim 8 , wherein the one or more KPIs comprise one or more of:

a network profit;

a transportation cost; and

an overhead cost.

14. The computer-implemented method of claim 8 , wherein calculating the range of feasible values comprises:

calculating, by the computer, a standard deviation and a mean of the predicted KPI value using a KPI lower bound and KPI upper bound of the selected risk range based on a base KPI value; and

when the standard deviation is non-zero, setting, by the computer, the KPI lower bound and the KPI upper bound equal to a Gaussian interval for the standard deviation, the mean, and the selected confidence interval.

15. A non-transitory computer-readable medium embodied with software, the software when executed by at least one server, the at least one server comprising a processor and memory:

requests, via a first asynchronous application programming interface call to one or more solvers of a solvers layer, to open a supply chain model of a supply chain network;

checks, via a second application programming interface call from the backend layer at regular intervals to the one or more solvers of the solvers layer, whether the one or more solvers are busy processing any request or are available to handle the request;

models, using a backend layer, the supply chain network as a supply chain planning problem, one or more key process indicators (KPIs) of the supply chain planning problem are based, at least in part, on the one or more input variables, wherein the backend layer uses one or more data update and retrieval Application Programming Interfaces (APIs) to trigger the one or more solvers to fetch data for the modelling;

models, via an application programming interface call from the backend layer to the solvers layer, an impact on the one or more KPIs from each of the one or more input variables at a selected confidence interval using a Bayesian optimization process; and

displays, using a UI layer, by one or more call back scripts that transmit requests to the solvers layer and by one or more python script modules, a visualization of a risk profile for the one or more KPIs, the visualization indicating a probability that an actual KPI value differs from a predicted KPI value.

16. The non-transitory computer-readable medium of claim 15 , the software when executed further:

receives a risk range selection, the risk range selection indicating a level of acceptable risk associated with the predicted KPI value;

calculates an optimal value for the one or more input variables and a range of feasible values for the one or more input variables, wherein the feasible values of the one or more input variables generate the predicted KPI value with a risk within the selected risk range; and

displays a visualization of the risk profile with the selected risk range.

17. The non-transitory computer-readable medium of claim 16 , the software when executed further:

calculates a standard deviation and a mean of the predicted KPI value using a KPI lower bound and KPI upper bound of the selected risk range based on a base KPI value; and

when the standard deviation is non-zero, sets the KPI lower bound and the KPI upper bound equal to a Gaussian interval for the standard deviation, the mean, and the selected confidence interval.

18. The non-transitory computer-readable medium of claim 15 , the software when executed further:

identifies sample points for building a surrogate model using an acquisition function; and

determines the sample points from the acquisition function by calculating input values having high uncertainties and input values having high derivative values.

19. The non-transitory computer-readable medium of claim 15 , the software when executed further:

builds a surrogate model by approximating a relationship between each of the one or more input variables and a resulting KPI value, wherein the surrogate model is constructed using the Bayesian optimization process.

20. The non-transitory computer-readable medium of claim 15 , wherein the visualization of the risk profile further comprises a visualization of a likelihood at a current input value, wherein the current input value is calculated as a difference of a cumulative distribution function from a Gaussian interval for a mean and standard deviation at an optimal input value and at a set sequential percentage value from the optimal input value.