CTNF 18/385,587 CTNF 100894 DETAILED ACTION This non-final office action is responsive to application 18/385,587 as submitted on 31 October 2023. Claim status is currently pending and under examination for claims 1-20 of which independent claims are 1 and 11. Notice of Pre-AIA or AIA Status 07-03-aia AIA 15-10-aia The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA. Claim Rejections - 35 USC § 101 07-04-01 AIA 07-04 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. Independent Claims 1 and 11 Step 2A Prong One: Does the claim recite an abstract idea, law of nature, or natural phenomenon? Yes, independent claim 1, under the broadest reasonable interpretation, recites the following limitations that are abstract ideas: identifying target time series data, a time range, and one or more features, wherein the one or more features may be categorical or numerical; (mental process) performing decomposition on the target time series data and numerical features, resulting in decomposed target time series data and decomposed numerical features; (mental process) performing linear regression based on the decomposed time series data, decomposed numerical features, and categorical features; (mental process and math) computing a residual based on the target time series data and a result of the linear regression; (mental process) determining a forecasted residual based on the residual; (mental process) and determining a multivariate time series forecast based on the results of the linear regression and the forecasted residual (mental process) The “identifying” step involves determining a time range and categorical/numerical features of time series data which amounts to no more than observations, evaluations, and judgments that can be performed in the human mind or with the use of a physical aid (e.g., pen and paper). The claim recites the step at a high degree of generality, thus the step is not required to have any specific level of complexity that would preclude the step from being mental processes. Therefore, the “identifying” step is considered to be mental processes, see MPEP § 2106.04(a)(2)(III). The “performing decomposition” step involves identifying isolated numerical features and trends from time series data which amounts to no more than observations, evaluations, and judgments that can be performed in the human mind or with the use of a physical aid (e.g., pen and paper). The claim recites the step of performing decomposition at a high degree of generality, thus the step is not required to have any specific level of complexity that would preclude the step from being mental processes. Therefore, the “performing decomposition” step is considered to be mental processes, see MPEP § 2106.04(a)(2)(III). The “performing linear regression” step involves using decomposed time series data and features to determine a prediction which represents a mathematical formula and amounts to no more than evaluations, observations, and judgments that can be performed in the human mind or with the use of a physical aid (e.g., pen and paper). The claim recites the step of performing linear regression at a high degree of generality, thus the step is not required to have any specific level of complexity that would preclude the step from being mental processes. Therefore, the “performing linear regression” step is considered to be a mathematical concept, see MPEP § 2106.04(a)(2)(I), and mental processes, see MPEP § 2106.04(a)(2)(III). The “computing” step involves determining a residual by using time series data and a result obtained from linear regression which amounts to no more than observations, evaluations, and judgments that can be performed in the human mind or with the use of a physical aid (e.g., pen and paper). The claim recites the step of computing a residual at a high degree of generality, thus the step is not required to have any specific level of complexity that would preclude the step from being mental processes. Therefore, the “computing” step is considered to be mental processes, see MPEP § 2106.04(a)(2)(III). The “determining a forecasted residual” step involves identifying a forecasted residual by using a computed residual which amounts to no more than observations, evaluations, and judgments that can be performed in the human mind or with the use of a physical aid (e.g., pen and paper). The claim recites the step of determining a forecasted residual at a high degree of generality, thus the step is not required to have any specific level of complexity that would preclude the step from being mental processes. Therefore, the “determining a forecasted residual” step is considered to be mental processes, see MPEP § 2106.04(a)(2)(III). The “determining a multivariate time series forecast” step involves identifying a time series forecast by using a result of linear regression and a forecasted residual which amounts to no more than observations, evaluations, and judgments that can be performed in the human mind or with the use of a physical aid (e.g., pen and paper). The claim recites the step of determining a multivariate time series forecast at a high degree of generality, thus the step is not required to have any specific level of complexity that would preclude the step from being mental processes. Therefore, the step is considered to be mental processes, see MPEP § 2106.04(a)(2)(III). Therefore, the independent claim recites a judicial exception. Independent claim 11 recites similar limitations corresponding to claim 1, therefore the same subject matter eligibility analysis is applied. Step 2A Prong Two: Does the claim recite additional elements that integrate the judicial exception into a practical application? No, the judicial exception recited above is not integrated into a practical application. The claims recite the following additional elements, but these additional elements are not sufficient to integrate the judicial exception into a practical application: a data warehouse arranged for storing time series data (MPEP § 2106.05(f) mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea) and one or more processors in communication with the data warehouse (MPEP § 2106.05(f) mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea) The remaining additional elements are recited at a high-level of generality such that they amount to no more than mere instructions to “apply” an exception using a generic component. Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea, see MPEP § 2106.05(f). Therefore, the above limitations do not integrate the judicial exception into a practical application. Step 2B: Does the claim recite additional elements that amount to significantly more than the judicial exception? No. The claims do not include additional elements that are sufficient for the claims to amount to significantly more than the judicial exception. In regards to the remaining additional elements, the limitations are recited so generically such that they amount to no more than mere instructions to “apply” the judicial exception on a computer using generic computer components. Mere instructions to apply a judicial exception cannot provide an inventive concept. See MPEP § 2106.05(f). Therefore, independent claims 1 and 11 are not patent eligible. Dependent Claims 2-10 and 12-20 The remaining dependent claims being rejected do not recite additional elements, whether considered individually or in combination, that are sufficient to integrate the judicial exception into a practical application or amount to significantly more than a judicial exception. Dependent Claim 2 recites the further limitations “wherein the decomposed target time series data is the target time series data with at least one of holiday or seasonality data removed.” The step is recited at a high-level of generality such that the limitations amount to no more than mere instructions to “apply” the judicial exception on a computer. They can also be viewed as nothing more than an attempt to generally link the use of the judicial exception to the technological environment of computers, see MPEP § 2106.05(f). Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea, see MPEP § 2106.05(f). The step does not integrate the judicial exception into a practical application and does not amount to significantly more. Dependent Claim 3 recites the further limitations “wherein the decomposed numerical features are the numerical features with at least one or holiday or seasonality data removed.” The step is recited at a high-level of generality such that the limitations amount to no more than mere instructions to “apply” the judicial exception on a computer. They can also be viewed as nothing more than an attempt to generally link the use of the judicial exception to the technological environment of computers, see MPEP § 2106.05(f). Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea, see MPEP § 2106.05(f). The step does not integrate the judicial exception into a practical application and does not amount to significantly more. Dependent Claim 4 recites the further limitations “wherein performing linear regression comprises assigning weights to each of the decomposed numerical features.” The step is recited at a high-level of generality such that the limitations amount to no more than mere instructions to “apply” the judicial exception on a computer. They can also be viewed as nothing more than an attempt to generally link the use of the judicial exception to the technological environment of computers, see MPEP § 2106.05(f). Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea, see MPEP § 2106.05(f). The step does not integrate the judicial exception into a practical application and does not amount to significantly more. Dependent Claim 5 recites the following limitations: calculating the matrix multiplication X' * X, where X is a traverse of X; (mental process and math) calculating an inverse of the matrix multiplication; (mental process and math) and multiplying the inverse with the target time series (mental process and math) The “calculating the matrix multiplication” step involves multiplying matrices, which represents a mathematical calculation and amounts to no more than evaluations, observations, and judgments that can be performed in the human mind or with the use of a physical aid (e.g., pen and paper). The claim recites the step of calculating a matrix multiplication at a high degree of generality, thus the step is not required to have any specific level of complexity that would preclude the step from being mental processes. Therefore, the step is considered to be a mathematical concept, see MPEP § 2106.04(a)(2)(I), and mental processes, see MPEP § 2106.04(a)(2)(III). The “calculating an inverse” step involves determining an inverse of a computed matrix multiplication, which represents a mathematical calculation and amounts to no more than evaluations, observations, and judgments that can be performed in the human mind or with the use of a physical aid (e.g., pen and paper). The claim recites the step of calculating an inverse at a high degree of generality, thus the step is not required to have any specific level of complexity that would preclude the step from being mental processes. Therefore, the step is considered to be a mathematical concept, see MPEP § 2106.04(a)(2)(I), and mental processes, see MPEP § 2106.04(a)(2)(III). The “multiplying the inverse” step involves determining a product from multiplying an inverse matrix with a time series, which represents a mathematical calculation and amounts to no more than evaluations, observations, and judgments that can be performed in the human mind or with the use of a physical aid (e.g., pen and paper). The claim recites the step at a high degree of generality, thus the step is not required to have any specific level of complexity that would preclude the step from being mental processes. Therefore, the “multiplying the inverse” step is considered to be a mathematical concept, see MPEP § 2106.04(a)(2)(I), and mental processes, see MPEP § 2106.04(a)(2)(III). This claim does not recite any non-abstract additional elements. Dependent Claim 6 recites the further limitations “wherein determining the forecasted residual comprises computing an autoregressive integrated moving average (ARIMA) model.” The step is recited at a high-level of generality such that the limitations amount to no more than mere instructions to “apply” the judicial exception on a computer. They can also be viewed as nothing more than an attempt to generally link the use of the judicial exception to the technological environment of computers, see MPEP § 2106.05(f). Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea, see MPEP § 2106.05(f). The step does not integrate the judicial exception into a practical application and does not amount to significantly more. Dependent Claim 7 recites the further limitations “wherein determining the multivariate time series forecast comprises summing the residual forecast with the result of the linear regression.” The step involves identifying a multivariate time series forecast by adding a residual forecast with a result of linear regression, which represents a mathematical calculation and amounts to no more than evaluations, observations, and judgments that can be performed in the human mind or with the use of a physical aid (e.g., pen and paper). The claim recites the step at a high degree of generality, thus the step is not required to have any specific level of complexity that would preclude the step from being mental processes. Therefore, the step is considered to be a mathematical concept, see MPEP § 2106.04(a)(2)(I), and mental processes, see MPEP § 2106.04(a)(2)(III). This claim does not recite any non-abstract additional elements. Dependent Claim 8 recites the further limitations “further comprising encoding the categorical data with numeric values prior to performing the linear regression.” The “encoding” step involves assigning numeric values to categorical data which amounts to no more than observations, evaluations, and judgments that can be performed in the human mind or with the use of a physical aid (e.g., pen and paper). The claim recites the step of encoding categorical data at a high degree of generality, thus the step is not required to have any specific level of complexity that would preclude the step from being mental processes. Therefore, the “encoding” step is considered to be mental processes, see MPEP § 2106.04(a)(2)(III). This claim does not recite any non-abstract additional elements. Dependent Claim 9 recites the further limitations “wherein the target time series data is stored in a data warehouse and accessed by the at least one processor from the data warehouse for the training using structured query language.” The limitation represents mere necessary data gathering and is recited at a high level of generality, thus adding insignificant extra-solution activity to the judicial exception - see MPEP § 2106.05(g). The extra-solution activity is a well-understood, routine and conventional (WURC) activity per MPEP § 2106.05(d)(II), “the courts have recognized the following computer functions as well-understood, routine, and conventional functions when they are claimed in a merely generic manner (e.g., at a high level of generality) or as insignificant extra-solution activity. i. Receiving or transmitting data over a network, e.g., using the Internet to gather data.” The limitation does not integrate the judicial exception into a practical application and does not amount to significantly more. Dependent Claim 10 recites the further limitations “further comprising forecasting the time series data using the multivariate forecasting model.” The step is recited at a high-level of generality such that the limitations amount to no more than mere instructions to “apply” the judicial exception on a computer. They can also be viewed as nothing more than an attempt to generally link the use of the judicial exception to the technological environment of computers, see MPEP § 2106.05(f). Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea, see MPEP § 2106.05(f). The step does not integrate the judicial exception into a practical application and does not amount to significantly more. Dependent claim 12 recites similar limitations corresponding to claim 2, therefore the same subject matter eligibility analysis is applied. Dependent claim 13 recites similar limitations corresponding to claim 3, therefore the same subject matter eligibility analysis is applied. Dependent claim 14 recites similar limitations corresponding to claim 4, therefore the same subject matter eligibility analysis is applied. Dependent claim 15 recites similar limitations corresponding to claim 5, therefore the same subject matter eligibility analysis is applied. Dependent claim 16 recites similar limitations corresponding to claim 6, therefore the same subject matter eligibility analysis is applied. Dependent claim 17 recites similar limitations corresponding to claim 7, therefore the same subject matter eligibility analysis is applied. Dependent claim 18 recites similar limitations corresponding to claim 8, therefore the same subject matter eligibility analysis is applied. Dependent claim 19 recites similar limitations corresponding to claim 9, therefore the same subject matter eligibility analysis is applied. Dependent claim 20 recites similar limitations corresponding to claim 10, therefore the same subject matter eligibility analysis is applied. Claim Rejections - 35 USC § 103 07-06 AIA 15-10-15 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. 07-20-aia AIA 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. 07-21-aia AIA Claim s 1-4, 6-7, 9-14, 16-17 and 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over Date (“Introduction to Regression with ARIMA Errors Model”) with WayBack Machine Publication date of 20 May 2022 having the link: https://web.archive.org/web/20220520095428/https://timeseriesreasoning.com/contents/regression-with-arima-errors-model/ in view of Abramson et al. (US 20190347670 A1), hereinafter Abramson. With respect to claim 1 , Date teaches: A method of training a multivariate forecasting model in a system … storing time series data and one or more processors …, the method comprising (Date discloses a SARIMAX model (‘multivariate forecasting model’) is trained (fit), “Let’s fit the SARIMAX model on the training data set (y_train, X_train) using the above parameters” (P. 15, Sec. “ Step 4 …” , ¶1). Date discloses “(S)ARIMA models are perfectly suited for forecasting time series data , and particularly for dealing with auto-correlated data. We apply this property of SARIMA models to model the auto-correlated residuals of the Linear Regression model after it is fitted to the time series data set ” (P. 5, Last Paragraph). Date discloses “We’ll design a SARIMAX model for a real world data set . The data set we’ll use contains hourly readings … Let’s begin by importing all the Python packages we will be using … We’ll use pandas … to load the data into a DataFrame” (P. 7, Sec. “ Using Python …” , ¶1-5). A computer is implied by using the pandas Python package to import a dataset, which further implies a system storing time series data and one or more processors.) : identifying target time series data, a time range, and one or more features (Date discloses a matrix of regression variables contains Temperature and Absolute Humidity variables (‘one or more features’), “The matrix of regression variables X will contain two variables: Temperature T Absolute Humidity AH We will use the Regression with (Seasonal) ARIMA errors i.e. the (S)ARIMAX model to predict hourly values of PT08_S4_NO2 using hourly measurements of Temperature and Absolute Humidity ” (P. 8, Sec. “Regression Strategy” , ¶2-3). Date discloses “Suppose your time series data set consists of a response variable and some regression variables . Suppose also that the regression variables are contained in a matrix X , and the response variable a.k.a. dependent variable is contained in a vector y . At each time step i , y takes some value y_i and there is a corresponding row vector x_i in X containing the values of all the regression variables at time step i " (P. 2, First Paragraph). Date discloses a figure on P. 2 (reproduced below) depicting a time series data set (‘target time series data’) consisting of regression variables (‘features’) and a response variable. The time series data is ordered by time step i in a range from y 1 to y n (older to newer, as represented by Arrow of Time), therefore identifying a time range and target time series data. PNG media_image1.png 601 1099 media_image1.png Greyscale ) , wherein the one or more features may be categorical or numerical (Date discloses a figure on P. 7 (reproduced below) depicting rows of an Air Quality data set and each variable’s corresponding values. Regression Variables T (temperature) and AH (Absolute Humidity) are represented as continuous values (decimals), therefore the regression variables are numerical. PNG media_image2.png 425 1620 media_image2.png Greyscale ) ; performing linear regression based on the … time series data, … numerical features … (Date discloses “We’ll now fit an Ordinary Least Squares Linear Regression model on the training data set and fetch it’s vector of residual errors. … In the expression below, we are saying that PT08_S4_NO2 is the dependent variable and T and AH are the regression variables . An intercept term is assumed by default. expr = 'PT08_S4_NO2 ~ T + AH'” (P. 9, Sec. “ STEP 2 … ”, ¶1-3). See (P. 2, First Paragraph) above describing a time series data set consists of regression variables T (temperature) and AH (Absolute Humidity).) ; computing a residual based on the target time series data and a result of the linear regression (See (P. 9, Sec. “ STEP 2 … ”, ¶1-3) above describing how a vector of residual errors is fetched (computed) from the Ordinary Least Squares Linear Regression Model that was fitted to the time series data set.) ; determining a forecasted residual based on the residual (Date discloses parameters of a SARIMAX model are estimated using the residuals of the Ordinary Least Squares Linear Regression Model (OLSR model), “We now begin the process of estimating the parameters of the SARIMA model on the OLSR model’s residual errors of regression ε ” (P. 10, Last Paragraph). Date discloses “( S)ARIMA models are perfectly suited for forecasting time series data, and particularly for dealing with auto-correlated data . We apply this property of SARIMA models to model the auto-correlated residuals of the Linear Regression model after it is fitted to the time series data set. The resulting model is known as Regression with Seasonal ARIMA Errors or SARIMAX for short. It can be loosely specified as follows: PNG media_image3.png 200 400 media_image3.png Greyscale If (p,d,q), (P,D,Q) and m are chosen correctly, one would expect the residual errors of the SARIMAX model to be not auto-correlated. ” (P. 5-6, Last Paragraph). A SARIMAX model is defined by y i = β ^ x i + ∈ i , where β ^ is a fitted linear regression model’s regression coefficients (see last figure on P. 2), x i is regression variables, and ∈ i is the modeled residuals obtained using ARIMA that are not auto-correlated. The residuals ∈ i are modeled (forecasted) using ARIMA and the auto-correlated residuals of the fitted OLSR model, therefore residuals ∈ i are forecasted residuals.) ; and determining a multivariate time series forecast based on the results of the linear regression and the forecasted residual (Date discloses a SARIMAX model is defined by y i = β ^ x i + ∈ i , where y i is a response (dependent) variable to be forecasted (see P. 2, First Paragraph above), β ^ is a fitted linear regression model’s regression coefficients (therefore results of linear regression), x i is regression variables, and ∈ i (‘forecasted residual’) is the modeled residuals obtained using ARIMA, see (P. 5-6, Last Paragraph above). Date discloses “The final step in our SARIMAX saga is using the chosen model to generate some predictions. We’ll ask the model to generate 24 out-of-sample predictions i.e. predict the value of the y (pollutant PT08_S4_NO2) for the next 24 hours beyond the end of the training data set . We’ll use the testing data set ( y_test , X_test ) which we had carved out in STEP-1. Recollect that the model has never seen the test data set during training … Call the get_forecast method to get the out of sample forecasts: predictions = sarimax_results.get_forecast(steps = 24, exog = X_test_minus_intercept[:24])” (P. 16-17, Sec. “ Step 5 …” , ¶1-3). Date discloses Temperature and Absolute Humidity variables are used to predict hourly values of PT08_S4_NO2 by using a SARIMAX model (see P. 8, Sec. “Regression Strategy” , ¶2-3), therefore a SARIMAX model is a multivariate time series model that uses multiple regression variables and time series to forecast a pollutant PT08_S4_NO2 (and therefore determining a multivariate time series forecast).) . However, Date does not teach a data warehouse and performing linear regression based on decomposed time series data, decomposed numerical features, and categorical features, which is taught by Abramson: a system comprising a data warehouse arranged for storing time series data and one or more processors in communication with the data warehouse (The Examiner interprets “data warehouse” according to its broadest reasonable interpretation (BRI) in view of the applicant’s specification (at [0004]). Accordingly, the Examiner interprets this term as encompassing a cloud-based server cluster as disclosed by Abramson below. A cloud-based server cluster is comprised of server devices, data storage, and routers that transmit and receive data in the form of SQL queries, therefore a cloud-based server cluster is a data management system (and therefore a data warehouse). Abramson discloses a cloud-based server cluster (‘data warehouse’), “FIG. 2 depicts a cloud-based server cluster 200 in accordance with example embodiments. In FIG. 2, operations of a computing device (e.g., computing device 100 of FIG. 1) may be distributed between server devices 202, data storage 204, and routers 206, all of which may be connected by local cluster network 208 … The server cluster 200 may, for example, be configured to execute (e.g., via computer processors of the server devices 202 thereof), in whole or in part, the method of FIG. 4 .” [0052]. See Figure 2 depicting a cloud-based server cluster comprised of server devices, data storage, and routers. Abramson discloses “ Server devices 202 may be configured to transmit data to and receive data from cluster data storage 204 . This transmission and retrieval may take the form of SQL queries” [0058]. Abramson discloses “ At step 402, the processor retrieves weather data from a first server . The first server may be a weather server (e.g., weather server 302 of FIG. 3) that is coupled to the processor via an electronic communications network (e.g., communication network(s) 306 of FIG. 3). The weather data may be stored on a weather database (e.g., weather database 308 of FIG. 3) of the first server” [0116]. See [0061] describing weather data is time series data.) , identifying target time series data … and one or more features, wherein the one or more features may be categorical or numerical (Abramson discloses weather and energy usage data are time series data (‘target time series data’), “predictively calculate future energy usage for the building via analysis of multiple time-series datastreams (e.g., including time-series weather data and time-series energy usage data) using an analytics engine 320” [0061]. Abramson discloses weekday/weekend operation tags and working day/holiday tags are derived variables (categorical variables) from weather and energy usage data, “the analytics engine 320 may execute a dataframe function to calculate derived variables and/or statistics from the energy usage data (e.g., energy usage statistics and/or energy usage variables) and the weather data (e.g., to calculate weather statistics and/or weather variables). For example, the derived variables may include , for each day for which data is available, one or more of: the change in energy consumption over 15 minute intervals, … a weekday or weekend operation tag for that day, a working day or holiday tag for that day ” [0070].) ; performing decomposition on the target time series data and numerical features, resulting in decomposed target time series data and decomposed numerical features (Abramson discloses weather and energy usage data are combined into an energy audit dataset, “ combining the weather data, utility energy usage data , and building-specific characteristic data into an energy audit dataset ” [0027]. Abramson discloses weather variables include numerical features (temperature, wind speed), “the weather database 308 may include time-series datastreams for each of a number of weather variables (e.g., which may include outdoor/ambient temperature, wind speed ” [0065]. Abramson discloses “pre-processing the energy audit dataset to generate a pre-processed energy audit dataset by generating a set of tags defining days for which full data of the weather data and the energy usage data is available, identifying and replacing missing values of the energy audit dataset, detecting and replacing anomalous data points by, performing time-series decomposition of the energy audit dataset to produce a decomposed time series, identifying outliers of the decomposed time series, and removing and replacing the identified outliers from the energy audit dataset,” [0029]. By performing time-series decomposition of the energy audit dataset to produce a decomposed time series (‘decomposed target time series data’), numerical variables (temperature, wind speed) from the energy audit dataset, must also be decomposed (therefore performing decomposition on the numerical features, resulting in decomposed numerical features).) ; performing linear regression based on the decomposed time series data, decomposed numerical features, and categorical features (See [0007] describing how input data is weather/energy usage data and a set of variables generated using a data frame function (see [0070] describing how weekday/weekend operation tags and working day/holiday tags are derived variables (categorical variables) generated from a data frame function). Abramson discloses input data (weather/energy usage data (energy audit dataset) and set of variables (categorical variables)) are used in linear regression, “the analytics engine 320 may first remove weekends and holidays from the input data to produce modified input data that includes modified time series weather data and modified time-series energy usage data . Next, the analytics engine 320 may apply a piecewise linear regression model to the modified time series energy usage data vs. the exterior temperature data of the modified time-series weather data to produce two linear regression trendlines , one for the heating season and one for the cooling season” [0099]. Abramson discloses the energy audit dataset is pre-processed to perform time-series decomposition to identify and remove outliers/anomalies from the resulting decomposed time series (‘decomposed time series data’), see [0029]. See Figure 4 (not reproduced) describing Steps 410 and 414. In Step 410, merged data (energy usage and weather data) is pre-processed to remove/correct anomalous data points. Then at Step 414, the pre-processed merged data is analyzed to produce analytics data. Since the energy audit dataset (input data) is decomposed first in a pre-processing step, the input data that has holidays removed must be decomposed time series data and decomposed numerical features used for linear regression.) . Abramson teaches a cloud-based server cluster that supports the transmission and retrieval of time series data and performing linear regression based on decomposed time series data and categorical variables is a known method in the art. Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the forecasting method of Date with the linear regression technique disclosed by Abramson to decompose time series data before performing linear regression. By decomposing time series data before performing linear regression, outliers and holiday data can be removed, thereby reducing noise and improving the accuracy of a linear regression model. With respect to claim 2 , the combined forecasting method of Date/Abramson teaches: The method of claim 1, wherein the decomposed target time series data is the target time series data with at least one of holiday or seasonality data removed (See [0007] describing how input data is weather/energy usage data. Abramson discloses holidays are removed from input data (weather/energy usage data (energy audit dataset)) before performing linear regression, see [0099] above. Abramson discloses the energy audit dataset is pre-processed to perform time-series decomposition to identify and remove outliers/anomalies from the resulting decomposed time series (‘decomposed time series data’), see [0029]. See Figure 4 (not reproduced) describing Steps 410 and 414. In Step 410, merged data (energy usage and weather data) is pre-processed to remove/correct anomalous data points. Then at Step 414, the pre-processed merged data is analyzed to produce analytics data. Since the energy audit dataset (input data) is decomposed first in a pre-processing step, the input data that has holidays removed must be decomposed time series data with holidays removed.) . Abramson teaches removing holidays from decomposed time series data is a known method in the art. Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined forecasting method of Date/Abramson with the holiday removal technique disclosed by Abramson to remove holiday data from decomposed time series data before performing linear regression. By removing holiday data from decomposed time series data before performing linear regression, holidays represented as outliers can be identified and removed, thereby creating a more accurate and less skewed regression line that represents the underlying trends between regression variables. With respect to claim 3 , the combined forecasting method of Date/Abramson teaches: The method of claim 1, wherein the decomposed numerical features are the numerical features with at least one or holiday or seasonality data removed (See [0007] describing how input data is weather/energy usage data. Abramson discloses holidays are removed from input data (weather/energy usage data (energy audit dataset)) before performing linear regression, see [0099] above. Abramson discloses weather variables from weather data include numerical features (temperature, wind speed), see [0065]. Abramson discloses the energy audit dataset is pre-processed to perform time-series decomposition to identify and remove outliers/anomalies from the resulting decomposed time series (decomposed time series data), see [0029]. See Figure 4 describing Steps 410 and 414. In Step 410, merged data (energy usage and weather data) is pre-processed to remove/correct anomalous data points. Then at Step 414, the pre-processed merged data is analyzed to produce analytics data. Since the energy audit dataset (input data) is decomposed first in a pre-processing step, the input data that has holidays removed must be decomposed time series data with holidays removed. By performing time-series decomposition of the energy audit dataset to produce decomposed time series data, numerical variables (temperature, wind speed) from the energy audit dataset, must also be decomposed (therefore removing holiday data from decomposed numerical features).) . Abramson teaches removing holidays from decomposed numerical features is a known method in the art. Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined forecasting method of Date/Abramson with the holiday removal technique disclosed by Abramson to remove holiday data from decomposed numerical features before performing linear regression. By removing holiday data from decomposed numerical features before performing linear regression, holidays represented as outliers can be identified and removed, thereby creating a more accurate and less skewed regression line that represents the underlying trends between regression variables. With respect to claim 4 , the combined forecasting method of Date/Abramson teaches: The method of claim 1, wherein performing linear regression comprises assigning weights to each of the decomposed numerical features (Date discloses “In the above model specification, β(cap) is an (m x 1) size vector storing the fitted model’s regression coefficients ” (P. 3, First Paragraph). See Last figure on P. 2 depicting a linear regression model defined as y = β ^ X + ∈ , where β ^ is the fitted linear regression model’s regression coefficients (‘weights’) assigned to regression variables X. Date discloses a figure on P. 7 (reproduced above) depicting regression Variables T (temperature) and AH (Absolute Humidity) are numerical features. Abramson discloses holidays are removed from input data (weather/energy usage data) before performing linear regression, see [0099] above. Abramson discloses weather variables from weather data include numerical features (temperature, wind speed), see [0065]. Abramson discloses the energy audit dataset (weather/energy usage data) is pre-processed to perform time-series decomposition to identify and remove outliers/anomalies from the resulting decomposed time series (decomposed time series data), see [0029]. The decomposed time series also includes decomposed numerical variables (temperature, wind speed) that are used for linear regression after holidays are removed, see explanation above.) . Abramson teaches using decomposed numerical variables in linear regression is a known method in the art. Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined forecasting method of Date/Abramson with the decomposed numerical variables disclosed by Abramson to use decomposed numerical features in linear regression. By using decomposed numerical features in linear regression, seasonal data can be isolated and removed, thereby allowing accurate underlying trends to be identified for forecasting and accurate decision-making. With respect to claim 6 , the combined forecasting method of Date/Abramson teaches: The method of claim 1, wherein determining the forecasted residual comprises computing an autoregressive integrated moving average (ARIMA) model (A SARIMAX model is defined by y i = β ^ x i + ∈ i , where β ^ is a fitted linear regression model’s regression coefficients, x i is regression variables, and ∈ i is the modeled residuals obtained using ARIMA that are not auto-correlated (see P. 5-6, Last Paragraph above). The residuals ∈ i are modeled (forecasted) using ARIMA and the auto-correlated residuals of the fitted OLSR model, therefore residuals ∈ i are forecasted residuals.) . With respect to claim 7 , the combined forecasting method of Date/Abramson teaches: The method of claim 1, wherein determining the multivariate time series forecast comprises summing the residual forecast with the result of the linear regression (Date discloses a figure on P. 6 (reproduced below) depicting an equation that defines a SARIMAX model. PNG media_image4.png 477 914 media_image4.png Greyscale A SARIMAX model is defined by y i = β ^ x i + ∈ i , where y i is a response (dependent) variable to be forecasted (‘multivariate time series forecast’), β ^ is a fitted linear regression model’s regression coefficients (therefore results of linear regression), x i is regression variables, and ∈ i (‘forecasted residual’) is the modeled residuals obtained using ARIMA, see (P. 5-6, Last Paragraph above). To determine y i (‘multivariate time series forecast’), β ^ x i and ∈ i are summed (therefore determining the multivariate time series forecast by summing the residual forecast with the result of the linear regression).) . With respect to claim 9 , the combined forecasting method of Date/Abramson teaches: The method of claim 1, wherein the target time series data is stored in a data warehouse and accessed by the at least one processor from the data warehouse for the training using structured query language (Abramson discloses a cloud-based server cluster (‘data warehouse’), “FIG. 2 depicts a cloud-based server cluster 200 in accordance with example embodiments. In FIG. 2, operations of a computing device (e.g., computing device 100 of FIG. 1) may be distributed between server devices 202, data storage 204, and routers 206, all of which may be connected by local cluster network 208 … The server cluster 200 may, for example, be configured to execute (e.g., via computer processors of the server devices 202 thereof), in whole or in part, the method of FIG. 4 .” [0052]. See Figure 2 depicting a cloud-based server cluster comprised of server devices, data storage, and routers. Abramson discloses server devices of a cloud-based server cluster can retrieve (access and store) data from the cluster’s data storage by using SQL queries, “ Server devices 202 may be configured to transmit data to and receive data from cluster data storage 204 . This transmission and retrieval may take the form of SQL queries” [0058]. Abramson discloses “ At step 402, the processor retrieves weather data from a first server . The first server may be a weather server (e.g., weather server 302 of FIG. 3) that is coupled to the processor via an electronic communications network (e.g., communication network(s) 306 of FIG. 3). The weather data may be stored on a weather database (e.g., weather database 308 of FIG. 3) of the first server” [0116]. See [0061] describing weather data is time series data. See [0099] describing time series weather data is used for linear regression.) . Abramson teaches server devices of a cloud-based server cluster using SQL queries to retrieve time series data for linear regression is a known method in the art. Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined forecasting method of Date/Abramson with the data retrieval technique disclosed by Abramson to use SQL queries to retrieve time series data from a data storage. By using SQL queries to retrieve time series data from a data storage, relevant time series data can be retrieved according to user queries and needs, thereby ensuring that the most relevant data can be used to train an accurate linear regression model. With respect to claim 10 , the combined forecasting method of Date/Abramson teaches: the method of claim 1, further comprising forecasting the time series data using the multivariate forecasting model ( Date discloses “The final step in our SARIMAX saga is using the chosen model to generate some predictions. We’ll ask the model to generate 24 out-of-sample predictions i.e. predict the value of the y (pollutant PT08_S4_NO2) for the next 24 hours beyond the end of the training data set . We’ll use the testing data set ( y_test , X_test ) ” (P. 16-17, Sec. “ Step 5 …” , ¶1-3). Date discloses Temperature and Absolute Humidity variables are used to predict hourly values of PT08_S4_NO2 by using a SARIMAX model (see P. 8, Sec. “Regression Strategy” , ¶2-3), therefore a SARIMAX model is a multivariate time series model that uses multiple regression variables and time series to forecast a pollutant PT08_S4_NO2 (and therefore forecasting time series data using the multivariate forecasting model). ) . With respect to claim 11 , the rejection of claim 1 is incorporated. The difference in scope being: A system for training a multivariate forecasting model, comprising … one or more processors configured to (Date discloses a SARIMAX model (‘multivariate forecasting model’) is trained (fit), “Let’s fit the SARIMAX model on the training data set (y_train, X_train) using the above parameters” (P. 15, Sec. “ Step 4 …” , ¶1). A computer is implied by fitting a SARIMAX model on training data, which further implies one or more processors.) . With respect to claim 12, the claim recites similar limitations corresponding to claim 2, therefore the same rationale of rejection is applicable. With respect to claim 13, the claim recites similar limitations corresponding to claim 3, therefore the same rationale of rejection is applicable. With respect to claim 14, the claim recites similar limitations corresponding to claim 4, therefore the same rationale of rejection is applicable. With respect to claim 16, the claim recites similar limitations corresponding to claim 6, therefore the same rationale of rejection is applicable. With respect to claim 17, the claim recites similar limitations corresponding to claim 7, therefore the same rationale of rejection is applicable. With respect to claim 19, the claim recites similar limitations corresponding to claim 9, therefore the same rationale of rejection is applicable. With respect to claim 20, the claim recites similar limitations corresponding to claim 10, therefore the same rationale of rejection is applicable . 07-21-aia AIA Claim s 5 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Date in view of Abramson, further in view of Lobelle et al. (“Detectability of an AMOC decline in current and projected climate changes”), hereinafter Lobelle . With respect to claim 5 , the combined forecasting method of Date/Abramson teaches The method of claim 4, however the combination does not teach calculating a matrix multiplication, which is taught by Lobelle: wherein assigning the weights comprises: calculating the matrix multiplication X' * X, where X is a traverse of X (Lobelle discloses “Generalized Least Squares (GLS) Once again, the starting point is the same linear regression as Equation 1, written in matrix notation as: PNG media_image5.png 134 1353 media_image5.png Greyscale ” (P. 4, Sec. 3.4, ¶1). Lobelle discloses “ V is a covariance matrix. In order to estimate the linear regression coefficient, the covariance matrix must be incorporated into the previous OLS estimate in Equation 2, as follows: ω ^ G L S = ( X ' V - 1 X ) - 1 X ' V - 1 Y , where X′ reflects the transpose of X (defined through the classical Euclidian scalar product)” (P. 4, Sec. 3.4, ¶2). Lobelle discloses Equation 9 on P. 4 (reproduced below) depicting an equation for Generalized Least Squares (GLS) used for estimating linear regression coefficient ω ^ G L S . PNG media_image6.png 334 1112 media_image6.png Greyscale ( X ' V - 1 X ) of GLS represents the product obtained from multiplying matrix X with matrix X’ (a traverse of matrix X), therefore the resulting product is matrix multiplication of X' * X.) ; calculating an inverse of the matrix multiplication (Lobelle discloses ( X ' V - 1 X ) - 1 of GLS represents an inverse of matrix multiplication ( X ' V - 1 X ) , see Equation 9 above.) ; and multiplying the inverse with the target time series (Lobelle discloses “ Y t are the data (annual means of AMOC transport …), 𝜔 is the trend magnitude … n is the sample size (i.e., corresponding to the total number of years of the time series Y t )” (P. 2-3, Sec. 3, Last Paragraph Lobelle discloses ( X ' V - 1 X ) - 1 of GLS is multiplied by Y (times series), see Equation 9 above.) . Lobelle teaches using Generalized Least Squares to multiply an inverse of matrix multiplication with time series data is a known method in the art. Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined forecasting method of Date/Abramson with the Generalized Least Squares equation disclosed by Lobelle to use Generalized Least Squares to estimate linear regression coefficients. By using Generalized Least Squares to estimate linear regression coefficients, optimal coefficients can be identified for a regression model, thereby improving linear regression accuracy. With respect to claim 15, the claim recites similar limitations corresponding to claim 5, therefore the same rationale of rejection is applicable . 07-21-aia AIA Claim s 8 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Date in view of Abramson, further in view of Francis et al. (“Prediction of engine demand with a data-driven approach”), hereinafter Francis . With respect to claim 8 , the combined forecasting method of Date/Abramson teaches The method of claim 1, however, the combination does not teach encoding categorical data with numeric values prior to performing linear regression, which is taught by Francis: further comprising encoding the categorical data with numeric values prior to performing the linear regression (Francis discloses “In linear regression, a linear equation is used to represent the relationship between a dependent variable and one or more explanatory variables . … An additive common-slope linear regression model with a dummy-variable regressor , as described by Fox 27 , is shown in equation (1). PNG media_image7.png 70 1392 media_image7.png Greyscale where α is the intercept for the baseline model, β is the coefficient for the quantitative explanatory variable X, γ is the coefficient for the dummy variable regressor D , and ε is the error. The dummy variable regressor D is coded such that D = 1 or D = 0 . For the engine sales data set, the dummy variable regressor allows the model to capture seasonality by effectively changing the intercept for each month” (P. 30, Sec. 3.1). Francis discloses “ Eleven qualitative dummy-variables (months) were used to account for seasonality . Each dummy variable was coded as 1 or 0. For example, a dummy variable representing the month of November would equal 1 if the sales of a data point occurred in November and 0, otherwise ” (P. 32-33, Sec. 4.1). 11 qualitative dummy variables (coded as 1 or 0) were used in linear regression to represent the occurrence of a sale in a given month, therefore encoding categorical data with numerical values prior to performing linear regression.) . Francis teaches using dummy variables to represent the occurrence of sales in a given month for linear regression is a known method in the art. Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined forecasting method of Date/Abramson with the dummy variables disclosed by Francis to convert categorical variables into numerical values. By converting categorical variables into numerical values, encoded qualitative data can be processed by mathematical formulas used in linear regression to make predictions, thereby ensuring relevant categorical variables can be used to make accurate and interpretable predictions. With respect to claim 18, the claim recites similar limitations corresponding to claim 8, therefore the same rationale of rejection is applicable. Conclusion Any inquiry concerning this communication or earlier communications from the examiner should be directed to PEDRO J MORALES whose telephone number is (571)272-6106. The examiner can normally be reached 8:30 AM - 6:00 PM. Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice. If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, MIRANDA M HUANG can be reached at (571)270-7092. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300. Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000. /PEDRO J MORALES/Examiner, Art Unit 2124 /VINCENT GONZALES/Primary Examiner, Art Unit 2124 Application/Control Number: 18/385,587 Page 2 Art Unit: 2124 Application/Control Number: 18/385,587 Page 3 Art Unit: 2124 Application/Control Number: 18/385,587 Page 4 Art Unit: 2124 Application/Control Number: 18/385,587 Page 5 Art Unit: 2124 Application/Control Number: 18/385,587 Page 6 Art Unit: 2124 Application/Control Number: 18/385,587 Page 7 Art Unit: 2124 Application/Control Number: 18/385,587 Page 8 Art Unit: 2124 Application/Control Number: 18/385,587 Page 9 Art Unit: 2124 Application/Control Number: 18/385,587 Page 10 Art Unit: 2124 Application/Control Number: 18/385,587 Page 11 Art Unit: 2124 Application/Control Number: 18/385,587 Page 12 Art Unit: 2124 Application/Control Number: 18/385,587 Page 13 Art Unit: 2124 Application/Control Number: 18/385,587 Page 14 Art Unit: 2124 Application/Control Number: 18/385,587 Page 15 Art Unit: 2124 Application/Control Number: 18/385,587 Page 16 Art Unit: 2124 Application/Control Number: 18/385,587 Page 17 Art Unit: 2124 Application/Control Number: 18/385,587 Page 18 Art Unit: 2124 Application/Control Number: 18/385,587 Page 19 Art Unit: 2124 Application/Control Number: 18/385,587 Page 20 Art Unit: 2124 Application/Control Number: 18/385,587 Page 21 Art Unit: 2124 Application/Control Number: 18/385,587 Page 22 Art Unit: 2124 Application/Control Number: 18/385,587 Page 23 Art Unit: 2124 Application/Control Number: 18/385,587 Page 24 Art Unit: 2124 Application/Control Number: 18/385,587 Page 25 Art Unit: 2124 Application/Control Number: 18/385,587 Page 26 Art Unit: 2124 Application/Control Number: 18/385,587 Page 27 Art Unit: 2124 Application/Control Number: 18/385,587 Page 28 Art Unit: 2124 Application/Control Number: 18/385,587 Page 29 Art Unit: 2124 Application/Control Number: 18/385,587 Page 30 Art Unit: 2124
