# US 2014/0195466 A1 · Integrated machine learning for a data management product

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2014/0195466 A1
- Document ID: US20140195466A1
- Status: Published application
- Kind: A1
- Application number: 14/150,327
- Filed: 2014-01-08
- Published: 2014-07-10
- Earliest claimed: 2013-01-08
- Assignees: PurePredictive, Inc.
- Applicants: PurePredictive, Inc.
- Inventors: Kelly D. Phillipps; Richard W. Wellman; Milind D. Zodge
- Web page: https://patentgrep.com/patent/US20140195466A1

## CPC

- G06N 20/20: Ensemble learning
- G05B 13/048: {using a predictor}
- G06F 16/00: Information retrieval; Database structures therefor; File system structures therefor
- G06N 20/00: Machine learning
- G06N 20/10: using kernel methods, e.g. support vector machines [SVM]
- G06Q 10/04: Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" (market predictions or forecasting for commercial activities G06Q30/0202)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR

## Abstract

Apparatuses, systems, methods, and computer program products are disclosed for machine learning in a data management product. The apparatus includes an input module, a learned function module, and a results module. The input module is configured to receive an analysis request for the data management product. The learned function module is configured to execute one or more machine learning ensembles to predict one or more unknown values for the data management product. The result module is configured to provide native access, within the data management product, to the one or more unknown values.

## Claims

### Claim 1

An apparatus for machine learning in a data management product, the apparatus comprising:
an input module configured to receive an analysis request for a data management product;
a learned function module configured to execute one or more machine learning ensembles to predict one or more unknown values for the data management product; and
a result module configured to provide native access, within the data management product, to the one or more unknown values.

### Claim 2 (depends on 1)

The apparatus of claim 1, wherein one or more of the input module, the learned function module, and the result module are integrated with the data management product as one or more of an add-on, a plug-in, and an extension.

### Claim 3 (depends on 1)

The apparatus of claim 1, wherein the results module is further configured to provide native access, within the data management product, to a confidence value corresponding to each of the one or more unknown values.

### Claim 4 (depends on 1)

The apparatus of claim 1, further comprising an ensemble factory module configured to generate the one or more machine learning ensembles using training data from the data management product, the one or more machine learning ensembles configured to predict the one or more unknown values in the data management product using workload data from the data management product.

### Claim 5 (depends on 1)

The apparatus of claim 1, wherein the results module provides native access through one or more of a native query language and a native interface of the data management product.

### Claim 6

A method for machine learning in a data management product, the method comprising:
receiving an analysis request for a data management product;
executing one or more machine learning ensembles to predict one or more unknown values for the data management product; and
providing native access, within the data management product, to one or more predicted values corresponding to the one or more unknown values.

### Claim 7 (depends on 6)

The method of claim 6, further comprising providing native access, within the data management product, to a confidence value corresponding to each of the one or more unknown values.

### Claim 8 (depends on 6)

The method of claim 6, further comprising modifying a data structure of the data management product to store the one or more predicted values.

### Claim 9 (depends on 6)

The method of claim 6, wherein providing native access includes providing access through one or more of a native query language and a native interface of the data management product.

### Claim 10 (depends on 6)

The method of claim 6, further comprising generating the one or more machine learning ensembles using training data from the data management product, the one or more machine learning ensembles configured to predict the one or more unknown values in the data management product using workload data from the data management product.

### Claim 11 (depends on 10)

The method of claim 10, further comprising receiving, by the machine learning module, a new ensemble request from the data management product, wherein generating the one or more machine learning ensembles comprises:
receiving the training data, the training data being associated with the new ensemble request;
generating a plurality of learned functions using the training data;
evaluating the plurality of learned function using test data from the training data; and
forming a machine learning ensemble comprising two or more learned functions based on the evaluation of the plurality of learned functions.

### Claim 12 (depends on 11)

The method of claim 11, wherein the plurality of learned functions are generated from multiple classes using the training data and the machine learning ensemble is formed by combining a subset of the plurality of learned functions from at least two classes of the multiple classes.

### Claim 13 (depends on 10)

The method of claim 10, wherein generating the one or more machine learning ensembles comprises:
generating a plurality of learned functions;
evaluating the plurality of learned function;
combining the plurality of learned functions in a plurality of learned function combinations;
evaluating the plurality of learned function combinations;
extending the plurality of learned function combinations with one or more layers, at least one of the layers comprising a probabilistic model;
evaluating the extended combined learned functions using test data;
selecting a subset of the extended combined learned functions;
synthesizing the selected subset of the extended combined learned functions with metadata generated by the evaluations; and
organizing the synthesized subset into a machine learning ensemble.

### Claim 14 (depends on 10)

The method of claim 10, wherein the one or more machine learning ensembles are generated automatically in response to one or more of a command and a query from a user of the data management product without further interaction with the user.

### Claim 15 (depends on 14)

The method of claim 14, further comprising regenerating the one or more machine learning ensembles in response to a trigger event.

### Claim 16 (depends on 15)

The method of claim 15, wherein the trigger event includes one or more of a change in data stored in the data management product, a manual update request, and a passing of a predetermined amount of time.

### Claim 17 (depends on 6)

The method of claim 6, wherein the data management product comprises a spreadsheet product.

### Claim 18 (depends on 6)

The method of claim 6, wherein the data management product comprises a database product that operates using a structured query language (SQL).

### Claim 19

A method for machine learning in a data management product, the method comprising:
providing a machine learning module in communication with a data management product;
transmitting a new ensemble request from the data management product to the machine learning module;
receiving training data from the data management product;
generating a plurality of learned functions using the training data;
forming a machine learning ensemble based on the plurality of learned functions; and
integrating the machine learning ensemble with the data management product to provide access to machine learning results within the data management product.

### Claim 20 (depends on 19)

The method of claim 19, wherein forming a machine learning ensemble based on the plurality of learned functions comprises:
evaluating the plurality of learned function;
combining the plurality of learned functions in a plurality of learned function combinations;
evaluating the plurality of learned function combinations;
extending the plurality of learned function combinations with one or more layers, at least one of the layers comprising a probabilistic model;
evaluating the extended combined learned functions using test data;
selecting a subset of the extended combined learned functions;
synthesizing the selected subset of the extended combined learned functions with metadata generated by the evaluations; and
organizing the synthesized subset into a machine learning ensemble.