# US 2020/0265071 A1 · Mixed initiative feature engineering

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

- Number: US 2020/0265071 A1
- Document ID: US20200265071A1
- Status: Published application
- Kind: A1
- Application number: 16/280,528
- Filed: 2019-02-20
- Published: 2020-08-20
- Earliest claimed: 2019-02-20
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Srinivasan Parthasarathy; Tejaswini Pedapati
- Web page: https://patentgrep.com/patent/US20200265071A1

## CPC

- G06F 9/541: {via adapters, e.g. between incompatible applications}
- G06F 16/287: {Visualization; Browsing}
- G06F 9/505: {considering the load}
- G06F 9/5072: {Grid computing}
- G06F 16/254: {Extract, transform and load [ETL] procedures, e.g. ETL data flows in data warehouses}
- G06N 20/00: Machine learning
- G06F 2209/5019: Workload prediction
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Computerized interactive feature visualization is carried out on a data set—a plurality of insight classes rank a plurality of features of the data set. Via a computerized user interface, user feedback is obtained based on the interactive feature visualization—a user selects and ranks a subset of the features. At least one transformation function is applied to at least one feature of the subset of features selected by the user, to automatically construct, with a computer, at least one additional feature for the data set. The data set with the at least one additional feature is a transformed data set. In some cases, a supervised task is carried out on the final data set; accuracy of a machine learning system implementing the at least one supervised task can be enhanced by the at least one additional feature, and/or a physical system can be controlled based on results of the at least one supervised task.

## Claims

### Claim 1

A method comprising:
carrying out computerized interactive feature visualization on a data set, wherein a plurality of insight classes rank a plurality of features of said data set;
via a computerized user interface, obtaining user feedback based on said interactive feature visualization, wherein a user selects and ranks a subset of said features; and
applying at least one transformation function to at least one feature of said subset of features selected by said user, to automatically construct, with a computer, at least one additional feature for said data set, said data set with said at least one additional feature comprising a transformed data set.

### Claim 2 (depends on 1)

The method of claim 1, further comprising iteratively repeating said steps of carrying out interactive feature visualization, obtaining user feedback, and applying at least one transformation function, on said transformed data set, until a final data set is obtained.

### Claim 3 (depends on 2)

The method of claim 2, wherein said data set is tabular and said features comprise columns.

### Claim 4 (depends on 2)

The method of claim 2, wherein said data set is in JavaScript Object Notation (JSON) and said features comprise attributes.

### Claim 5 (depends on 2)

The method of claim 2, wherein said applying of said at least one transformation function comprises applying a univariate data transformation function with said selected features as arguments.

### Claim 6 (depends on 2)

The method of claim 2, wherein said applying of said at least one transformation function comprises applying a multivariate data transformation function with said selected features as arguments.

### Claim 7 (depends on 2)

The method of claim 2, wherein said applying of said at least one transformation function comprises applying a multivariate data transformation function to arguments comprising at least one of said selected features and at least one existing one of said features.

### Claim 8 (depends on 2)

The method of claim 2, further comprising carrying out at least one supervised task on said final data set.

### Claim 9 (depends on 8)

The method of claim 8, wherein said at least one supervised task comprises electricity demand prediction, further comprising at least one of:
causing additional electrical generation capacity to be activated, based on said prediction indicating increased demand; and
causing current electrical generation capacity to be reduced, based on said prediction indicating decreased demand.

### Claim 10 (depends on 8)

The method of claim 8, wherein said at least one supervised task comprises computer resources demand prediction, further comprising at least one of:
causing additional computer resource capacity to be activated, based on said prediction indicating increased demand; and
causing current computer resource capacity to be reduced, based on said prediction indicating decreased demand.

### Claim 11 (depends on 8)

The method of claim 8, wherein accuracy of a machine learning system implementing said at least one supervised task is enhanced by said at least one additional feature.

### Claim 12 (depends on 2)

The method of claim 2, further comprising carrying out at least one unsupervised task on said final data set.

### Claim 13 (depends on 2)

The method of claim 2, further comprising carrying out data imputation on said final data set.

### Claim 14 (depends on 2)

The method of claim 2, wherein, in said applying of said at least one transformation function, said at least one transformation function comprises a user-defined function in a database query language.

### Claim 15

A non-transitory computer readable medium comprising computer executable instructions which when executed by a computer cause the computer to perform a method of:
carrying out computerized interactive feature visualization on a data set, wherein a plurality of insight classes rank a plurality of features of said data set;
via a computerized user interface, obtaining user feedback based on said interactive feature visualization, wherein a user selects and ranks a subset of said features; and
applying at least one transformation function to at least one feature of said subset of features selected by said user, to automatically construct, with said computer, at least one additional feature for said data set, said data set with said at least one additional feature comprising a transformed data set.

### Claim 16 (depends on 15)

The non-transitory computer readable medium of claim 15, wherein the method performed by the computer implementing the instructions further comprises iteratively repeating said steps of carrying out interactive feature visualization, obtaining user feedback, and applying at least one transformation function, on said transformed data set, until a final data set is obtained.

### Claim 17 (depends on 16)

The non-transitory computer readable medium of claim 16, wherein the method performed by the computer implementing the instructions further comprises:
carrying out at least one supervised task on said final data set, said at least one supervised task comprising demand prediction for a physical system; and
at least one of:
causing additional physical capacity of said physical system to be activated, based on said prediction indicating increased demand, by sending a first signal over a network interface; and
causing current physical capacity of said physical system to be reduced, based on said prediction indicating decreased demand, by sending a second signal over said network interface.

### Claim 18

An apparatus comprising:
a memory; and
at least one processor, coupled to said memory, and operative to:
carry out computerized interactive feature visualization on a data set, wherein a plurality of insight classes rank a plurality of features of said data set;
implement a computerized user interface, which obtains user feedback based on said interactive feature visualization, wherein a user selects and ranks a subset of said features; and
apply at least one transformation function to at least one feature of said subset of features selected by said user, to automatically construct at least one additional feature for said data set, said data set with said at least one additional feature comprising a transformed data set.

### Claim 19 (depends on 18)

The apparatus of claim 18, wherein the at least one processor is further operative to iteratively repeat said carrying out interactive feature visualization, obtaining user feedback, and applying at least one transformation function, on said transformed data set, until a final data set is obtained.

### Claim 20 (depends on 19)

The apparatus of claim 19, further comprising a network interface coupled to said memory and said at least one processor, wherein said at least one processor is further operative to:
carry out at least one supervised task on said final data set, said at least one supervised task comprising demand prediction for a physical system; and
at least one of:
cause additional physical capacity of said physical system to be activated, based on said prediction indicating increased demand, by sending a first signal over said network interface; and
cause current physical capacity of said physical system to be reduced, based on said prediction indicating decreased demand, by sending a second signal over said network interface.
