System, method for deploying computing infrastructure, and method for identifying customers at risk of revenue change

1. A method for identifying a customer at risk of revenue change and identifying individual erosion risk factors of a portfolio of contracts, said method being executed by a processor, the method comprising:

constructing, as executed by the processor, a set of examples of client accounts;

specifying a set of candidate risk drivers of the client accounts;

computing, as executed by the processor, descriptive numeric features that capture erosion risk factors from each of the set of candidate risk drivers during a time window of interest;

forming a mathematical model, as executed by the processor, to compute a risk score of contract termination and individual risk scores associated with each of said erosion risk factors;

computing, as executed by the processor, a set of descriptive numeric features for the portfolio of contracts that is to be tracked based on the formed mathematical model;

applying the formed mathematical model, as executed by the processor, to the set of descriptive numeric features;

computing, as executed by the processor, a score indicating a risk exposure of an account in the portfolio of service contracts;

computing, as executed by the processor, individual risk factor scores indicating a contribution of each of the risk factors to the computed score; and

sorting the portfolio, as executed by the processor, to determine one of a client's risk, a client's financial pressures, a customer's unhappiness level, and how much of a client's revenue is exposed to high risk.

2. The method of claim 1 , wherein the examples comprise negative client accounts including accounts that have been terminated, re-scoped, or a contract has been renegotiated with another company.

3. The method of claim 1 , wherein the set of erosion risk factors comprise financial and business performance metrics, news-based metrics on significant changes in client's company, metrics that characterize client satisfaction with provider's quality of service and service delivery, metrics that characterize provider's price, and event-based metrics that characterize if a client had a history of contract termination.

4. The method of claim 1 , wherein, in a case of a negative action, the time-window comprises a period preceding contract termination or a contract change.

5. The method of claim 1 , wherein the mathematical model is formed from applying a statistical, machine learning operation to said set of examples to be optimized to produce a best forecast against these examples.

6. The method of claim 1 , wherein the mathematical model is formed from applying an optimization approach to the set of examples to be optimized to produce a best forecast against these examples.

7. The method of claim 1 , wherein the set of descriptive numeric features are extracted during the time window preceding the date of interest for which the risk score is to be computed.

8. The method of claim 1 , further comprising sorting, as executed by the processor, the portfolio to determine at least one of a client's risk, a client's financial pressures, a customer's unhappiness level, and how much of a client's revenue is exposed to high risk.

9. The method of claim 1 , further comprising identifying, as executed by the processor, combinations of descriptive numeric features that cause a significant change in the risk score indicating the risk exposure of the account in the portfolio of service contracts.

10. The method of claim 1 , wherein each one of the individual risk factor scores is multiplied by a determined weight factor.

11. The method of claim 1 , wherein the examples comprise positive client accounts.

12. A method for iteratively identifying a customer at risk of revenue change based on and identifying individual erosion risk factors of a portfolio of service contracts, said method being executed by a processor, the method comprising:

observing a series of training samples, as executed by the processor, each of the training samples having a plurality of training sample features;

identifying a series of variables, as executed by the processor, from the series of training samples;

iteratively splitting the series of samples, as executed by the processor, according to each of the variables;

determining an impact of each of the variables, as executed by the processor;

identifying combinations of variables, as executed by the processor, that result in a significant change in parameter values; and

sorting the portfolio, as executed by the processor, to determine one of a client's risk, a client's financial pressures, a customer's unhappiness level, and how much of a client's revenue is exposed to high risk.