Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Status of Claims
Claims 1 – 20 are pending and examined herein.
Claims 1 – 10, 19 – 20 are rejected under 35 U.S.C. 112(b).
Claims 1 – 20 are rejected under 35 U.S.C. 101.
Claims 1 – 20 are rejected under 35 U.S.C. 103.
Specification
The disclosure is objected to because of the following informalities:
Reference 150 used to refer “neural network training module” and “neural network module” in [0038]. 110 should be used to refer “neural network module”.
Reference 210 used to refer “neural network” instead of “neural network module” in [0044].
Reference 110 used to refer “neural network” instead of “neural network module” in [0050].
Appropriate correction is required.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 19 – 20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 19 recites the limitation “current the first list” in line 36. It is unclear whether the limitation is referring back to “a current first list” or “a first list” previously recited before line 36 of claim 19. There is insufficient antecedent basis for this limitation in the claim. For the examination purposes, “current the first list” will refer to the “a current first list” previously stated in claim 19.
Claim 20 is dependent on claim 19. It does not resolve the issue of indefiniteness and are rejected with the same rationale.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1 - 20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
MPEP § 2109(III) sets out steps for evaluating whether a claim is drawn to patent-eligible subject matter. The analysis of claims 1 – 20, in accordance with these steps, follows.
Step 1 Analysis:
Step 1 is to determine whether the claim is directed to a statutory category (process, machine, manufacture, or composition of matter.
Claims 1 – 10 are directed to a method, meaning that it is directed to the statutory category of process. Claims 11 – 18 are directed to a method, which is also the statutory category of process. Claims 19 – 20 are directed to a system, which can be an article of machine.
Step 2A Prong One, Step 2A Prong Two, and Step 2B Analysis:
Step 2A Prong One asks if the claim recites a judicial exception (abstract idea, law of nature, or natural phenomenon). If the claim recites a judicial exception, analysis proceeds to Step 2A Prong Two, which asks if the claim recites additional elements that integrate the abstract idea into a practical application. If the claim does not integrate the judicial exception, analysis proceeds to Step 2B, which asks if the claim amounts to significantly more than the judicial exception. If the claim does not amount to significantly more than the judicial exception, the claim is not eligible subject matter under 35 U.S.C. 101.
Regarding claim 1, the following claim elements are abstract ideas:
identifies and labels one or more features from the customer data to produce a portion of the featured-labeled data; (This is practical to perform in the human mind under its broadest reasonable interpretation aside from the recitation of generic computer components or by a human using a pen and paper.)
produce at least one list that predicts first customers predicted to leave the enterprise and second customers predicted to stay with enterprise when provided an incentive; (This is practical to perform in the human mind under its broadest reasonable interpretation aside from the recitation of generic computer components or by a human using a pen and paper.)
The following claim elements are additional elements which, taken alone or in combination with the other additional elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
obtaining customer data associated with customers of an enterprise; (This is mere data gathering, an insignificant extra solution activity, which is a well-understood, routine conventional activity. It does not integrate the judicial exception into a practical application. See MPEP § 2106.05(d). Therefore, this does not amount to significantly more than the judicial exception.)
processing classification machine-learning models (MLMs) on the customer data producing feature-labeled data, each classification MLM (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
processing a prediction MLM on the feature-labeled data to (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
and providing the at least one list to one or more of an enterprise interface and an enterprise promotion system (This is mere transmitting data, which is a well-understood, routine conventional activity. It does not integrate the judicial exception into a practical application. See MPEP § 2106.05(d). Therefore, this does not amount to significantly more than the judicial exception.)
to mitigate customer attrition with respect to the second customers via the incentive. (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
Regarding claim 2, the rejection of claim 1 is incorporated herein. Further, claim 2 recites the following additional ideas:
iterating to the obtaining at preconfigured intervals of time (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
Regarding claim 3, the rejection of claim 1 is incorporated herein. Further, claim 3 recites the following abstract idea:
processing a Recency, Frequency, Monetary (RFM) analyzer on the customer data to produce a second list of customers predicted by the RFM analyzer to leave the enterprise. (Running RFM algorithm is practical to perform in the human mind under its broadest reasonable interpretation aside from the recitation of generic computer components or by a human using a pen and paper. In addition, running algorithm could fall under a mathematical concept.)
Claim 3 does not recite additional elements.
Regarding claim 4, the rejection of claim 3 is incorporated herein. Further, claim 4 recites the following abstract idea:
calculating evaluation metrics including accuracy and precision rates for the at least one list and the second list and an F1 accuracy metric for the at least one first list in view of actual subsequently observed results associated with specific customers that left the enterprise. (Calculating such metrics like accuracy and precision rates is merely mathematical calculation, which is mathematical concept.)
Claim 4 does not recite additional elements.
Regarding claim 5, the rejection of claim 4 is incorporated herein. Further, claim 5 recites the following abstract idea:
flagging customer data associated with inaccurate predictions produced by the prediction MLM in the customer data as training data based on comparing the at least one list and the second list in view of the subsequently observed results. (This is practical to perform in the human mind under its broadest reasonable interpretation aside from the recitation of generic computer components or by a human using a pen and paper.)
Claim 5 does not recite additional elements.
Regarding claim 6, the rejection of claim 5 is incorporated herein. Further, claim 6 recites the following additional ideas:
training the prediction MLM using the flagged customer data and the subsequently observed results to improve the F1 accuracy metric of the prediction MLM. (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
Regarding claim 7, the rejection of claim 6 is incorporated herein. Further, claim 7 recites the following additional ideas:
iterating to the obtaining at preconfigured intervals of time. (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
Regarding claim 8, the rejection of claim 1 is incorporated herein. Further, claim 8 recites the following additional ideas:
wherein obtaining further includes obtaining the customer data from a plurality of data stores associated with the enterprise. (This is mere data gathering, an insignificant extra solution activity, which is a well-understood, routine conventional activity. It does not integrate the judicial exception into a practical application. See MPEP § 2106.05(d). Therefore, this does not amount to significantly more than the judicial exception.)
Regarding claim 9, the rejection of claim 8 is incorporated herein. Further, claim 9 recites the following abstract idea:
obtaining further includes separating the customer data into sets of data, each set associated with a specific customer. (This is practical to perform in the human mind under its broadest reasonable interpretation aside from the recitation of generic computer components or by a human using a pen and paper.)
Claim 9 does not recite additional elements.
Regarding claim 10, the rejection of claim 9 is incorporated herein. Further, claim 10 recites the following abstract idea:
wherein separating further includes separating the customer data into the sets of data using non-personal identification information associated with each customer to anonymize the sets of data associated with the customers. (This is practical to perform in the human mind under its broadest reasonable interpretation aside from the recitation of generic computer components or by a human using a pen and paper.)
Claim 10 does not recite additional elements.
Regarding claim 11, the following claim elements are abstract ideas:
labeling customer data for customer of an enterprise with features created labeled data; (This is practical to perform in the human mind under its broadest reasonable interpretation aside from the recitation of generic computer components or by a human using a pen and paper.)
labeling the current customer data with the features creating current labeled data; (This is practical to perform in the human mind under its broadest reasonable interpretation aside from the recitation of generic computer components or by a human using a pen and paper.)
Claim 11 further recites the following additional elements:
training a machine-learning model (MLM) to use the labeled data as input and labeled labeled actual results associated with customers leaving the enterprise as an expected output of the MLM; (This is mere data gathering and outputting, an insignificant extra solution activity, which does not integrate the judicial exception into a practical application. The broadest reasonable interpretation of this claim is storing information in memory, which is a well-understood, routine conventional activity. See MPEP § 2106.05(d)(II)(iv). Therefore, this does not amount to significantly more than the judicial exception.)
obtaining current customer data for the enterprise; (This is mere data gathering, an insignificant extra solution activity, which is a well-understood, routine conventional activity. It does not integrate the judicial exception into a practical application. See MPEP § 2106.05(d). Therefore, this does not amount to significantly more than the judicial exception.)
providing the current labeled data as input to the MLM and receiving as output a first list of customers predicted to leave the enterprise and a second list of customers predicted to stay with the enterprise when provided an incentive from the enterprise; (This is mere data gathering and outputting, an insignificant extra solution activity, which does not integrate the judicial exception into a practical application. The broadest reasonable interpretation of this claim is storing information in memory, which is a well-understood, routine conventional activity. See MPEP § 2106.05(d)(II)(iv). Therefore, this does not amount to significantly more than the judicial exception.)
providing the first list and the second list to the enterprise. (This is mere transmitting data, which is a well-understood, routine conventional activity. It does not integrate the judicial exception into a practical application. See MPEP § 2106.05(d). Therefore, this does not amount to significantly more than the judicial exception.)
Regarding claim 12, the rejection of claim 11 is incorporated herein. Further, claim 12 recites the following abstract idea:
comparing a third list produced by a Recency, Frequency, Monetary (RFM) analyzer against the first list and the second list in view of actual observed results for specific customer that left the enterprise. (This is practical to perform in the human mind under its broadest reasonable interpretation aside from the recitation of generic computer components or by a human using a pen and paper.)
Claim 12 does not recite additional elements.
Regarding claim 13, the rejection of claim 12 is incorporated herein. Further, claim 13 recites the following abstract idea:
identifying select current labeled data associated with inaccurate predictions of the MLM in the first list or the second list. (This is practical to perform in the human mind under its broadest reasonable interpretation aside from the recitation of generic computer components or by a human using a pen and paper.)
Claim 13 does not recite additional elements.
Regarding claim 14, the rejection of claim 13 is incorporated herein. Further, claim 14 recites the following additional ideas:
re-training the MLM using the select current labeled data causing the MLM to update after the re- training. (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
Regarding claim 15, the rejection of claim 14 is incorporated herein. Further, claim 15 recites the following additional ideas:
c (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
Regarding claim 16, the rejection of claim 11 is incorporated herein. Further, claim 16 recites the following additional ideas:
wherein labeling the customer data further includes processing the customer data through classification MLMs to obtain the labeled customer data. (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
Regarding claim 17, the rejection of claim 16 is incorporated herein. Further, claim 17 recites the following additional ideas:
wherein processing further includes processing first classification MLMs in parallel against the customer data and merging output produced by the first classification into an intermediate labeled data. (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
Regarding claim 18, the rejection of claim 17 is incorporated herein. Further, claim 18 recites the following additional ideas:
wherein processing further includes processing at least one second classification MLM on the intermediate labeled data to obtain the labeled customer data. (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
Regarding claim 19, the following claim elements are abstract ideas:
comparing current the first list, the current second list, and the third list against actual observed results where specific customers left the enterprise and identifying in the current intermediate customer data portions associated with inaccurate predictions made by the attrition prediction MLM; (This is practical to perform in the human mind under its broadest reasonable interpretation aside from the recitation of generic computer components or by a human using a pen and paper.)
Claim 19 further recites the following additional elements:
a cloud server comprising at least one processor and a non-transitory computer-readable storage medium, (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
training classification machine learning models (MLMs) to identify features from customer data of customers associated with an enterprise; (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
merging labeled customer data into intermediate customer data; (This is mere transmitting data, which is a well-understood, routine conventional activity. It does not integrate the judicial exception into a practical application. See MPEP § 2106.05(d). Therefore, this does not amount to significantly more than the judicial exception.)
training an attrition prediction MLM to take as input the intermediate customer data and produce as output a first list of customers predicted to leave the enterprise and a second list of customers predicted to stay with the enterprise when provided an incentive; (This is mere data gathering and outputting, an insignificant extra solution activity, which does not integrate the judicial exception into a practical application. The broadest reasonable interpretation of this claim is storing information in memory, which is a well-understood, routine conventional activity. See MPEP § 2106.05(d)(II)(iv). Therefore, this does not amount to significantly more than the judicial exception.)
at predefined intervals of time obtaining updated customer data from data stores of the enterprise; during each interval: (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
processing the classification MLMs to obtain current labeled feature data; (This is mere data gathering, an insignificant extra solution activity, which is a well-understood, routine conventional activity. It does not integrate the judicial exception into a practical application. See MPEP § 2106.05(d). Therefore, this does not amount to significantly more than the judicial exception.)
merging the current labeled featured data into current intermediate customer data; (This is mere transmitting data, which is a well-understood, routine conventional activity. It does not integrate the judicial exception into a practical application. See MPEP § 2106.05(d). Therefore, this does not amount to significantly more than the judicial exception.)
processing the attrition prediction MLM to obtain a current first list of customers and a currency second list of customers; (This is mere data gathering, an insignificant extra solution activity, which is a well-understood, routine conventional activity. It does not integrate the judicial exception into a practical application. See MPEP § 2106.05(d). Therefore, this does not amount to significantly more than the judicial exception.)
using an application programming interface (API) to provide the current first list and the current second list to an enterprise interface and to an enterprise promotion system; (This is mere transmitting data, which is a well-understood, routine conventional activity. It does not integrate the judicial exception into a practical application. See MPEP § 2106.05(d). Therefore, this does not amount to significantly more than the judicial exception.)
processing a Recency, Frequency, Monetary (RFM) analyzer against the updated customer data and obtaining a third list of customer predicted by the RFM analyzer to leave the enterprise; (This is mere data gathering and outputting, an insignificant extra solution activity, which does not integrate the judicial exception into a practical application. The broadest reasonable interpretation of this claim is storing information in memory, which is a well-understood, routine conventional activity. See MPEP § 2106.05(d)(II)(iv). Therefore, this does not amount to significantly more than the judicial exception.)
and re-training the attrition prediction MLM on the current intermediate customer data portions as a continuous feedback loop to improve an F1 accuracy metric of the attrition prediction MLM in providing the current first list and the current second list. (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
Regarding claim 20, the rejection of claim 19 is incorporated herein. Further, claim 20 recites the following additional ideas:
wherein the operations associated with the using the API further includes rendering the fist list and the second list within a dashboard screen associated with the enterprise interface. (This falls under mere instructions to apply abstract idea on a generic computer. See MPEP § 2106.05(f). Therefore, this does not amount to significantly more than the judicial exception.)
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1, 8 are rejected under 35 U.S.C. 103 as being unpatentable over Sotela et al. (U.S. Pub. 2015/0310336 A1) in view of Dickinson et al. (U.S. Pub. 2009/0276289 A1).
Regarding Claim 1, Sotela teaches
A method, comprising: obtaining customer data associated with customers of an enterprise; ([0006] of Sotela states “The platform may be configured to receive customer activity data. The platform may then compute features associated with the customer activity data. These features are then inputted into a machine learning model used for predicting customer churn. Finally, the platform may then provide a report indicating customer churn predictions. The platform may be trained in a training phase prior to entering a prediction phase.” [0033] of Sotela states “Platform 100 may be deployed on with a telecommunications service provider's network. In this way, platform 100 may have access to network customer networks 1-N through which it may access and retrieve customer activity data. In turn, platform 100 may use the customer activity data to perform the churn predictions detailed in this disclosure. The result of the calculations may be provided to user 105 (e.g., telecommunications operator).”)
processing classification machine-learning models (MLMs) on the customer data producing feature-labeled data, each classification MLM identifies and labels one or more features from the customer data to produce a portion of the featured-labeled data; ([0006] of Sotela states “The platform may then compute features associated with the customer activity data. These features are then inputted into a machine learning model used for predicting customer churn.” [0042] of Sotela states “In some embodiments, a random forest algorithm may be employed by platform 100. A random forest algorithm may be composed of hundred or even several thousands of different decision trees. Each decision tree may be generated from a random selection of a subset of m predictors or input features using a sample of the training set.” Classification ML model (random forest) operate on customer data. The model “compute features associated with the customer activity data.)
processing a prediction MLM on the feature-labeled data to produce at least one list that predicts first customers predicted to leave the enterprise and ([0006] of Sotela states “Embodiments of the present disclosure may provide a platform configured to forecast customer churn in a prepaid or postpaid telecommunication network. The platform may be configured to receive customer activity data. The platform may then compute features associated with the customer activity data. These features are then inputted into a machine learning model used for predicting customer churn. Finally, the platform may then provide a report indicating customer churn predictions. The platform may be trained in a training phase prior to entering a prediction phase.” [0050] of Sotela states “For the prediction phase, customer activity data may first be encoded into the current features set. Then these features may be propagated into the predictive model to generate the predictions.” [0174] of Sotela states “FIG. 5 shows results of the Receiver Operator Curve (ROC) churner's prediction for each of the eight months. The output of the predictive model is a score (between 0 and 1) that indicates the likelihood of the customer in churning.”)
and providing the at least one list to one or more of an enterprise interface ([0006] of Sotela states “The platform may be configured to receive customer activity data. The platform may then compute features associated with the customer activity data. These features are then inputted into a machine learning model used for predicting customer churn. Finally, the platform may then provide a report indicating customer churn predictions. The platform may be trained in a training phase prior to entering a prediction phase.” [0033] of Sotela states “Platform 100 may be deployed on with a telecommunications service provider's network. In this way, platform 100 may have access to network customer networks 1-N through which it may access and retrieve customer activity data. In turn, platform 100 may use the customer activity data to perform the churn predictions detailed in this disclosure. The result of the calculations may be provided to user 105 (e.g., telecommunications operator).”)
Sotela does not explicitly teach
second customers predicted to stay with enterprise when provided an incentive;
an enterprise promotion system to mitigate customer attrition with respect to the second customers via the incentive.
However, Dickinson teaches that
second customers predicted to stay with enterprise when provided an incentive; ([0014] of Dickinson states “The system and method for customer retention looks to a set of customers. Historical data from at least one store may be received. This data includes historical transaction data of old customers. Likewise, recent customer data may be received from the at least one store. This data includes transactions for the present customers.” [0017] of Dickinson states “From the risk factors, a loss model may be generated. In some embodiments, the loss model may be calibrated by discounting the frequency risk factor of the attriters after date of attrition.” [0018] of Dickinson states “The loss model may be used, in conjunction with current transaction data, to generate the likelihood of loss for each of the current customers. This likelihood of loss for each customer may then be reported.” [0019] of Dickinson states “At least one retention measure may be generated for each customer by comparing the customer's transactions to the loss model and the risk factors. The retention measures may be outputted to the stores, and may be sent to a price optimization system. Likewise, the retention measures may be validated by applying them, calculating actual customer loss and comparing the actual customer loss to the loss model.” [0203] of Dickinson states “The process then progresses to step 2902 where customer specific retention initiatives are generated. These initiatives may be generated for customers which are considered “at risk” for attrition. These measures may include personalized advertisements, personalized promotions or other personal treatment, or change in product assortment to meet customer's needs.” [0205] of Dickinson states “The initiatives are ranked at step 2906 by their relative costs and benefits. Thus, the most cost efficient retention initiatives may be ranked highest.” Customers with high churn or loss score are customers predicted to leave and subset of at-risk customers for which Dickinson generates retention initiatives are customers predicted to stay with incentive.)
an enterprise promotion system to mitigate customer attrition with respect to the second customers via the incentive. ([0013] of Dickinson states “In particular, the system and methods relies upon a highly predictive measure of customer likelihood of loss. Such systems are useful for providing businesses with an advanced competitive tool to greatly reduce customer loss in a cost efficient manner.” [0018] of Dickinson states “The loss model may be used, in conjunction with current transaction data, to generate the likelihood of loss for each of the current customers. This likelihood of loss for each customer may then be reported.” [0019] of Dickinson states “At least one retention measure may be generated for each customer by comparing the customer's transactions to the loss model and the risk factors. The retention measures may be outputted to the stores, and may be sent to a price optimization system. Likewise, the retention measures may be validated by applying them, calculating actual customer loss and comparing the actual customer loss to the loss model.” [0203] of Dickinson states “The process then progresses to step 2902 where customer specific retention initiatives are generated. These initiatives may be generated for customers which are considered “at risk” for attrition. These measures may include personalized advertisements, personalized promotions or other personal treatment” [0205] of Dickinson states “In some embodiments, all retention measures of the ranked retention measures may be compared to a threshold for their cost benefit in order to select measures for the retention scheme.” Sotela provides a churn prediction report to an operator via a platform UI(enterprise interface) and Dickinson uses a customer loss predictor to generate retention measures and personalized promotions to reduce customer loss(corresponding to enterprise promotion system using incentive to mitigate attrition))
It would have been obvious to one with ordinary skill in the art before the effective filing date of the invention to combine the teachings of Sotela and Dickinson. Sotela teaches a churn prediction platform that computes features from customer activity data and produces customer churn likelihoods. Dickinson teaches using customer likelihood of loss scores to generate and deliver targeted retention measures to mitigate customer attrition. One with ordinary skill in the art would be motivated to incorporate the teachings of Dickinson into that of Sotela to use Sotela’s churn prediction with a promotion/retention system to automate targeted incentives, improve the overall effectiveness and output of churn mitigation in a predictable way.
Regarding claim 8, the rejection of claim 1 is incorporated herein. Furthermore, the combination of Sotela and Dickinson teaches
wherein obtaining further includes obtaining the customer data from a plurality of data stores associated with the enterprise. ([0069] of Dickinson states “The data 152 regarding customer purchasing may be provided from the Stores 124 to the Customer Retention Engine 150 (step 229). Customer data 152 may include purchasing frequency, items purchased and some identifier linking the purchase to the particular customer.” [0145] of Dickinson states “The Customer Data 152 may be provided from the Stores 124 to the Customer Data Analyzer 2102 and Loss Modeling Engine 2104.” Multiple stores are all enterprise data source holding customer information and these data are provided from the stores.)
Claims 9, 10 are rejected under 35 U.S.C. 103 as being unpatentable Sotela et al. (U.S. Pub. 2015/0310336 A1) in view of Dickinson et al. (U.S. Pub. 2009/0276289 A1), further in view of Bagul et al. (U.S. Pub. 2022/0058513).
Regarding claim 9, the rejection of claim 8 is incorporated herein. The combination of Sotela and Dickinson teaches
each set associated with a specific customer. ([0014] of Dickinson states “Likewise, recent customer data may be received from the at least one store. This data includes transactions for the present customers.” [0015] of Dickinson states “The transactions need to be linked to each customer. This linking may utilize any of payment identifiers, loyalty program identifiers, registry identifiers, or biometric identifiers. If there are conflicts between the identifiers, the conflict may be resolved in favor of the most accurate identifier.”)
The combination does not explicitly teach
obtaining further includes separating the customer data into sets of data,
However, Bagul teaches
obtaining further includes separating the customer data into sets of data, (Pg. 351 A. Data Description of Bagul states “7. CustomerID: Customer number. Nominal, unique integral number assigned to each customer… This process also involved creation of two more datasets using aggregation which are customer aggregate data and invoice aggregate data” Linking transactions to customers and aggregate by customer ID to create sets of data with a specific customer )
It would have been obvious to one with ordinary skill in the art before the effective filing date of the invention to combine the teachings of Sotela, Dickinson, and Bagul. Sotela teaches a churn prediction platform that computes features from customer activity data and produces customer churn likelihoods. Dickinson teaches using customer likelihood of loss scores to generate and deliver targeted retention measures to mitigate customer attrition. Bagul teaches organizing datasets into customer aggregate datasets keyed by non-personal identifiers and performing RFM analysis on those datasets. One with ordinary skill in the art would be motivated to incorporate the teachings of Bagul into the combination of Sotela, Dickinson to consistently group customer records while anonymizing personal details. Therefore, it allows easier data integration across multiple enterprise data stores and address privacy concerns in a predictable way.
Regarding claim 10, the rejection of claim 9 is incorporated herein. Furthermore, the combination of Sotela, Dickinson, and Bagul teaches
wherein separating further includes separating the customer data into the sets of data using non-personal identification information associated with each customer to anonymize the sets of data associated with the customers. ([0014] of Dickinson states “Likewise, recent customer data may be received from the at least one store. This data includes transactions for the present customers.” [0015] of Dickinson states “The transactions need to be linked to each customer. This linking may utilize any of payment identifiers, loyalty program identifiers, registry identifiers, or biometric identifiers. If there are conflicts between the identifiers, the conflict may be resolved in favor of the most accurate identifier.” Pg. 351 A. Data Description of Bagul states “7. CustomerID: Customer number. Nominal, unique integral number assigned to each customer… This process also involved creation of two more datasets using aggregation which are customer aggregate data and invoice aggregate data” Using non-personal identification information to organize customer datasets and these are effectively anonymized from the model as only those IDs are provided.)
Claim 2 is rejected under 35 U.S.C. 103 as being unpatentable Sotela et al. (U.S. Pub. 2015/0310336 A1) in view of Dickinson et al. (U.S. Pub. 2009/0276289 A1), further in view of Aleksandrova (NPL: “Application of Machine Learning for Churn Prediction Based on Transactional Data (RFM Analysis)”).
Regarding claim 2, the rejection of claim 1 is incorporated herein. The combination of Sotela and Dickinson teaches
iterating to the obtaining ([0015] of Sotela states “FIG. 5 is a chart showing a receiver operating curve of the prediction results for eight different months” Sotela shows the system is run repeatedly to keep obtaining data and predicting)
The combination doesn’t explicitly teach
at preconfigured intervals of time
However, Aleksandrova teaches
at preconfigured intervals of time (Pg. 1 Abstract of Aleksandrova states “RFM scores are calculated for every customer for a period of 6 months before the end date of examination. The target value for prediction models is a churn metric indicating whether the customer has made a transaction in the next 6 months following the RFM analysis or not. Several machine learning algorithms has been applied such as Two-Class Boosted Decision Trees, Two-Class Neural Networks, Two-Class Decision Jungle, Two-Class SVM and Two-Class Logistic Regression” Pg. 2 Methodology of Aleksandrova states “Regarding the purpose of the paper a split period has been chosen to separate transactions made before and after that period. This period could be different depending on the specifics of the subject area.”)
It would have been obvious to one with ordinary skill in the art before the effective filing date of the invention to combine the teachings of Sotela, Dickinson, and Aleksandrova. Sotela teaches a churn prediction platform that computes features from customer activity data and produces customer churn likelihoods. Dickinson teaches using customer likelihood of loss scores to generate and deliver targeted retention measures to mitigate customer attrition. Aleksandrova teaches using pre-configured time interval with RFM features and churn labels to train and evaluate churn prediction models. One with ordinary skill in the art would be motivated to incorporate the teachings of Aleksandrova into the combination of Sotela, Dickinson so churn prediction and retention workflows run at preconfigured intervals that matches business planning cycles, improving stability and smooth operation of the churn system in a predictable way.
Claims 3 – 7, 11 – 17, 19 – 20 are rejected under 35 U.S.C. 103 as being unpatentable Sotela et al. (U.S. Pub. 2015/0310336 A1) in view of Dickinson et al. (U.S. Pub. 2009/0276289 A1), Bagul et al. (U.S. Pub. 2022/0058513), further in view of Aleksandrova (NPL: “Application of Machine Learning for Churn Prediction Based on Transactional Data (RFM Analysis)”).
Regarding claim 3, the rejection of claim 1 is incorporated herein. The combination of Sotela, Dickinson, and Aleksandrova teaches
produce a second list of customers predicted by the RFM analyzer to leave the enterprise. (Pg.1 Abstract of Aleksandrova states “The main goal of the current study is to propose a combination of RFM analysis and machine learning algorithms for churn prediction based on mainly transactional data… RFM scores are calculated for every customer for a period of 6 months before the end date of examination. The target value for prediction models is a churn metric indicating whether the customer has made a transaction in the next 6 months following the RFM analysis or not. Several machine learning algorithms has been applied such as Two-Class Boosted Decision Trees, Two-Class Neural Networks, Two-Class Decision Jungle, Two-Class SVM and Two-Class Logistic Regression”)
The combination doesn’t explicitly teach
processing a Recency, Frequency, Monetary (RFM) analyzer on the customer data to
However, Bagul teaches that
processing a Recency, Frequency, Monetary (RFM) analyzer on the customer data to (Pg. 351 D. RFM Technique of Bagul states “RFM (Recency, Frequency, Monetary) analysis is a marketing model for customer segmentation. It is based on customer behavior. It groups customers based on their transactional history that is how recently, how often and how much did they buy. RFM helps divide customers into various clusters to identify customers who are more likely to discontinue the business relationship.” Bagul shows how RFM analysis is used on customer data to identify customers more likely to discontinue the business relationship. Together, they teach running an RFM analyzer on customer data to get RFM scores and identify customers predicted to churn.)
It would have been obvious to one with ordinary skill in the art before the effective filing date of the invention to combine the teachings of Sotela, Dickinson, Bagul, and Aleksandrova. Sotela teaches a churn prediction platform that computes features from customer activity data and produces customer churn likelihoods. Dickinson teaches using customer likelihood of loss scores to generate and deliver targeted retention measures to mitigate customer attrition. Bagul teaches organizing datasets into customer aggregate datasets keyed by non-personal identifiers and performing RFM analysis on those datasets. Aleksandrova teaches using RFM with machine learning models for churn prediction, evaluating models using metrics such as accuracy, precision, recall, F1 and AUC, and refining the models based on the evaluation results and time-windowed churn labels. One with ordinary skill in the art would be motivated to incorporate the teachings of Aleksandrova into the combination of Sotela, Dickinson, Bagul to improve predictive performance metrics and provide more robust, interpretable, and periodically updated churn system, implemented on cloud hosted software that displays churn lists and retention lists to enterprise dashboard and retention stratagem system in a predictable way.
Regarding claim 4, the rejection of claim 3 is incorporated herein. The combination of Sotela, Dickinson, Bagul, and Aleksandrova teaches
calculating evaluation metrics including accuracy and precision rates for the at least one list and the second list and an F1 accuracy metric for the at least one first list in view of actual subsequently observed results associated with specific customers that left the enterprise. (Pg.1 Abstract of Aleksandrova states “RFM scores are calculated for every customer for a period of 6 months before the end date of examination. The target value for prediction models is a churn metric indicating whether the customer has made a transaction in the next 6 months following the RFM analysis or not.” Pg.4 Results of Aleksandrova states “After training the models, a scoring and evaluation has been made with results showed on table 1. For all models in the table the threshold is 0.5, positive label: Inactive (e.g. the customer didn’t make a transaction), negative label: Active (e.g. the customer made a transaction)… The distribution of the target variable is highly uneven with 19% from customers making purchases after the split period and 81% churned, e.g. those who didn’t make purchases. Regarding this when splitting data for the training and test set a stratified option was chosen depending on the values of the target variable. The uneven distribution suggests to use Recall, Precision and F1 score, along with AUC for evaluating the models.” Compare predictions whether customer churn following the RFM analysis by checking transaction record in the next 6 months and compute accuracy, precision, f1.)
Regarding claim 5, the rejection of claim 4 is incorporated herein. The combination of Sotela, Dickinson, Bagul, and Aleksandrova teaches
flagging customer data associated with inaccurate predictions produced by the prediction MLM in the customer data as training data based on comparing the at least one list and the second list in view of the subsequently observed results. (Pg.1 Abstract of Aleksandrova states “RFM scores are calculated for every customer for a period of 6 months before the end date of examination. The target value for prediction models is a churn metric indicating whether the customer has made a transaction in the next 6 months following the RFM analysis or not.” Pg.4 Results of Aleksandrova states “The distribution of the target variable is highly uneven with 19% from customers making purchases after the split period and 81% churned, e.g. those who didn’t make purchases. Regarding this when splitting data for the training and test set a stratified option was chosen depending on the values of the target variable. The uneven distribution suggests to use Recall, Precision and F1 score, along with AUC for evaluating the models.” [0018] of Dickinson states “The loss model may be used, in conjunction with current transaction data, to generate the likelihood of loss for each of the current customers. This likelihood of loss for each customer may then be reported.” [0019] of Dickinson states “At least one retention measure may be generated for each customer by comparing the customer's transactions to the loss model and the risk factors. The retention measures may be outputted to the stores, and may be sent to a price optimization system. Likewise, the retention measures may be validated by applying them, calculating actual customer loss and comparing the actual customer loss to the loss model.” When computing performance metrics in Aleksandrova, it inherently identifies which customers were incorrectly predicted (false negatives and false positives) based on a comparison of model output to subsequently observed results.)
Regarding claim 6, the rejection of claim 5 is incorporated herein. The combination of Sotela, Dickinson, Bagul, and Aleksandrova teaches
training the prediction MLM using the flagged customer data and the subsequently observed results to improve the F1 accuracy metric of the prediction MLM. (Pg.4 Results of Aleksandrova states “The distribution of the target variable is highly uneven with 19% from customers making purchases after the split period and 81% churned, e.g. those who didn’t make purchases. Regarding this when splitting data for the training and test set a stratified option was chosen depending on the values of the target variable. The uneven distribution suggests to use Recall, Precision and F1 score, along with AUC for evaluating the models.” Pg. 7 Conclusion of Aleksandrova states “However, evaluation metrics as AUC, Recall, Precision and F1 score are better after selection of additional input variables which in the case study happen to be count of objects and customer type. By adding other relevant input variables depending on the subject area data scientists can refine the prediction process resulting in more accurate forecast of customer churn” Abstract of Dickinson states “The loss model may be used, in conjunction with current transaction data, to generate the likelihood of loss for each of the current customers, which may then be reported. Retention measures may be generated for each customer by comparing the customer's transactions to the loss model and the risk factors… the retention measures may be validated by comparing actua