# US 2001/0020284 A1 · Method and device for analyzing data

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

- Number: US 2001/0020284 A1
- Document ID: US20010020284A1
- Status: Published application
- Kind: A1
- Application number: 09/769,528
- Filed: 2001-01-26
- Published: 2001-09-06
- Earliest claimed: 2000-02-18
- Assignees: Fujitsu Limited
- Inventors: Hidetaka Tsuda; Hidehiro Shirai
- Web page: https://patentgrep.com/patent/US20010020284A1

## CPC

- G06F 11/008: {Reliability or availability analysis}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A rule present between a plurality of data values is extracted through data mining to output the rule and the information showing the reliability of the rule. A set-division evaluation-value S-ratio and a t-value showing the clearness of division when dividing a set constituted of a plurality of data values into two subsets are defined as the information showing the reliability of a rule. S-ratio=((S1+S2)/2)/S0 wherein, S0 is sum of squares of purpose variables of set before divided, S1, S2 are sums of squares of purpose variables of each subset after divided.

## Claims

### Claim 1

A data analyzing device comprising:
an original data group constituted of a plurality of original data values;
a data processing section for extracting a rule present between the original data values and generating a reliability-information-provided rule file to which the information showing the reliability of the rule is added; and
an analyzing tool for analyzing the original data values in accordance with the rule file.

### Claim 2 (depends on 1)

The data analyzing device according to claim 1, wherein
said data processing section clarifies the causality between and the influential degree of record items of the original data group and extracting a rule including hidden information or regularity through a statistical method.

### Claim 3 (depends on 1)

The data analyzing device according to claim 1, wherein
said data processing section extracts a rule including a guideline for items to be analyzed by the analyzing tool or combinations of the items in accordance with a data mining technique.

### Claim 4 (depends on 1)

The data analyzing device according to claim 1, wherein
said data processing section obtains a set-division evaluation value showing the clearness of division when dividing a set constituted of a plurality of original data values into two subsets as the information showing the reliability of the rule.

### Claim 5 (depends on 4)

The data analyzing device according to claim 4, wherein
the original data includes purpose variables purposing the search of fluctuation causes or fluctuation patterns, and
the set-division evaluation value is a value S-ratio obtained as,
S-ratio ((S1+S2)/2)S0,
wherein, S0 denotes the sum of squares of purpose variables before a set constituted of a plurality of data values is divided into two subsets and S1 and S2 respectively denote the sum of squares of purpose variables of subsets after divided.

### Claim 6 (depends on 4)

The data analyzing device according to claim 4, wherein
the original data includes purpose variables purposing the search of fluctuation causes or fluctuation patterns, and
the set-division evaluation value has a value t obtained as, 2 t = [LeftBracketingBar] X1 _ - X2 _ [RightBracketingBar] S1 + S2 N1 + N2 - 2 × ( 1 N1 + 1 N2 ) ⁢ ⁢ or t = [LeftBracketingBar] X1 _ - X2 _ [RightBracketingBar] S1 N1 2 + S2 N2 ′ ,
wherein, S1 and S2 respectively denote the sum of squares of purpose variables of sets obtained by dividing a set constituted of a plurality of original data values into two subsets, N1 and N2 respectively denote the number of factors of each subset after divided, and {overscore (X1)} and {overscore (X2)} respectively denote the average of subsets after divided.

### Claim 7 (depends on 1)

The data analyzing device according to claim 1, wherein
said data processing section obtains the information including factors or conditions from a secondary factor or condition downward as the information showing the reliability of the rule.

### Claim 8 (depends on 1)

The data analyzing device according to claim 1 further comprising:
a front data processing section for extracting a combination of optional data species or optional data items from the original data group.

### Claim 9 (depends on 1)

A data analyzing device according to claim 1 which analyzes a yield-deterioration factor in a fabrication process,
wherein the original data includes purpose variables purposing the search of fluctuation causes or fluctuation patterns and explanation variables for explaining the fluctuation of the purpose variables,
the purpose variables are fabrication yields and the explanation variables are variables including the history of a device used, test results, design information, and measured data.

### Claim 10

A data analyzing method comprising the steps of:
extracting a rule present between a plurality of original data values, generating a reliability-information-provided rule file to which the information showing the reliability of the rule is added, and outputting the information showing the reliability together with the rule; and
analyzing the original data values in accordance with the rule file.

### Claim 11 (depends on 10)

The data analyzing method according to claim 10, wherein the causality between or influential degree of items of each record in the original data values is clarified and a rule including hidden information or regularity is extracted in accordance with a statistical method.

### Claim 12 (depends on 10)

The data analyzing method according to claim 10, wherein a rule including an index for items to be analyzed or a combination of them when analyzing the original data is extracted in accordance with a data mining technique.

### Claim 13 (depends on 10)

The data analyzing method according to claim 10, wherein information showing the reliability of the rule is a set-division evaluation value showing the clearness of division when dividing a set constituted of a plurality of original data values into two subsets.

### Claim 14 (depends on 13)

The data analyzing method according to claim 13, wherein
the original data includes purpose variables purposing the search of fluctuation causes or fluctuation patterns, and
the set-division evaluation value is a value of S-ratio obtained as,
S-ratio=((S1+S2)/2)/S0,
wherein, S0 denotes the sum of squares of purpose variables before dividing a set constituted of a plurality of original data values into two subsets and S1 and S2 respectively denote the sum of squares of purpose variables of each subset after divided.

### Claim 15 (depends on 13)

The data analyzing method according to claim 13, wherein
the original data includes purpose variables purposing the search of fluctuation causes or fluctuation patterns, and
the set-division evaluation value is a value t obtained by, 3 t = [LeftBracketingBar] X1 _ - X2 _ [RightBracketingBar] S1 + S2 N1 + N2 - 2 × ( 1 N1 + 1 N2 ) ⁢ ⁢ or t = [LeftBracketingBar] X1 _ - X2 _ [RightBracketingBar] S1 N1 2 + S2 N2 ′ ,
wherein, S1 and S2 respectively denote the sum of squares of purpose variables of sets obtained by dividing a set constituted of a plurality of original data values into two subsets, N1 and N2 respectively denote the number of factors of each subset after divided, and {overscore (X1)} and {overscore (X2)} respectively denote the average of subsets after divided.

### Claim 16 (depends on 10)

The data analyzing method according to claim 10, wherein
information showing the reliability of the rule includes factors or conditions from a secondary factor or condition downward.

### Claim 17 (depends on 10)

The data analyzing method according to claim 10, wherein a combination of optional data species or optional data items is previously extracted from the original data values as the data to be processed to extract a rule.

### Claim 18 (depends on 10)

The data analyzing method according to claim 10, wherein
to analyze yield deterioration factors in a fabrication process,
the original data includes purpose variables purposing the search of fluctuation causes or fluctuation patterns and explanation variables for explaining the fluctuation of the purpose variables,
the purpose variables are fabrication yields and the explanation variables are variables including the history of a device used, test results, design information, and measured data.

### Claim 19 (depends on 10)

The data analyzing method according to claim 10, wherein each record in the original data has a plurality of item included in a plurality of different types of item groups as purpose variables purposing the search of fluctuation causes or fluctuation patterns, and explanation variables for explaining the fluctuation of the purpose variables; and analyzes a set included in or not included in a watched node in a regression tree diagram based on the extracted rule.

### Claim 20 (depends on 10)

The data analyzing method according to claim 10, wherein each record in the original data has purpose variables purposing the search of fluctuation causes or fluctuation patterns and a plurality of items included in a plurality of different types of item groups as explanation variables for explaining the fluctuation of the purpose variables and analyzes a set included or not included in a watched node in a regression tree diagram according the extracted rule.

### Claim 21

A computer readable medium for storing instructions, which when executed by a computer, causes the computer to perform the steps of:
extracting a rule present between a plurality of original data values, generating a reliability-information-provided rule file to which the information showing the reliability of the rule is added, and outputting the information showing the reliability together with the rule; and
analyzing the original data values in accordance with the rule file.
