Expert system for statistical design of experiments

Claim 2

A computer implemented method of designing a robust experiment to be carried out by a user, the method comprising the steps of: a. inputting information to the computer, the information including: i. a response variable, ii. factors to be varied in the experiment, iii. factor properties including indications representing which of the factors are noise factors and which of the factors are control factors; b. generating an experiment design in accordance with the information input to the computer in step a.; c. constructing two mathematical models, the first mathematical model containing terms representing hypothetical effects of all of the factors and the factor interactions on the response variable, the second mathematical model containing terms representing hypothetical effects of the control factors and control factor interactions on the response variable; d. evaluating the experiment design generated in step b. by utilizing the information input in step a. and the two mathematical models constructed in step c. to compute values of statistical properties for terms in each of the two mathematical models; e. communicating the values computed in step d. to the user; f. generating a redesigned experiment by performing steps b. through e. using user changes to the information originally input in step a., until the user is satisfied with the values communicated in step e.; g. creating a layout sheet specifying individual trials of the redesigned experiment to be performed by the user; h. performing an analysis of experimental data obtained by the user performing the trials of the redesigned experiment specified by the layout sheet created in step g.; and i. communicating a summary of the analysis performed in step h. to the user.

Claim 3

A computer implemented method of designing an experiment to be carried out by a user for studying model response, the steps of the method comprising: a. inputting information to the computer, the information including: i. a response variable, ii. a model type to be generated, iii. factors to be varied in the experiment, and iv. factor properties; b. generating an experiment design in accordance with the model type input in step a.; c. constructing a mathematical model containing terms representing hypothetical effects of the factors and interactions between the factors on the response variable; d. evaluating the experiment design generated in step b. by utilizing the information input in step a. and the mathematical model constructed in step c. to compute values of statistical properties for terms in the mathematical mode; e. communicating the values computed instep d. to the user; f. generating a redesigned experiment by performing steps b. through e. using user changes to the information originally input in step a. until the user is satisfied wit the values communicated in step e., wherein the user changes are suggested by a set of expert system forward chaining rules; g. creating a layout sheet specifying individual trials of the redesigned experiment to be performed by the user; h. performing an analysis of experimental data obtained by the user performing the trials of the redesigned experiment specified by the layout sheet created in step g.; and i. communicating a summary of the analysis performed in step h. to the user.

Claim 4

A computer implemented method of designing an experiment to be carried out by a user for studying model response, the steps of the method comprising: a. inputting information to the computer, the information including: i. a response variable, ii. a model type to be generated, iii. factors to be varied in the experiment, and iv. factor properties; b. generating an experiment design in accordance with the model type input in step a.; c. constructing a mathematical model containing terms representing hypothetical effects of the factors and interactions between the factors on the response variable; d. evaluating the experiment design generated in step b. by utilizing the information input in step a. and the mathematical model constructed in step c. to compute values of statistical properties for terms in the mathematical mode; e. communicating the values computed instep d. to the user; f. generating a redesigned experiment by performing steps b. through e. using user changes to the information originally input in step a. until the user is satisfied wit the values communicated in step e., wherein the user changes are suggested by a set of expert system forward chaining rules; g. creating a layout sheet specifying individual trials of the redesigned experiment to be performed by the user; h. performing an analysis of experimental data obtained by the user performing the trials of the redesigned experiment specified by the layout sheet created in step g.; and i. communicating a summary of the analysis performed in step h. to the user.

Claim 5

A computer implemented method of designing a robust experiment to be carried out by a user, the method comprising the steps of: a. inputting information to the computer, the information including: i. a response variable, ii. factors to be varied in the experiment, and iii. factor properties including indications representing which of the factors are noise factors and which of the factors are control factors; b. generating an experiment design in accordance with the information input to the computer in step a.; c. constructing two mathematical models, the first mathematical model containing terms representing hypothetical effects of all of the factors and the interactions between the factors on the response variable, the second mathematical model containing terms representing hypothetical effects of the control factors and interactions between control factors on the response variable; d. evaluating the experiment design generated in step b. by utilizing the information input in step a. and the two mathematical models constructed in step c. to compute values of statistical properties for terms in each of the two mathematical models; e. communicating the values computed in step d. to the user; f. generating a redesigned experiment by performing steps b. through e. using user changes to the information originally input in step a., until the user is satisfied with the values communicated in step e., wherein the user changes are suggested by a set of expert system forward chaining rules; g. creating a layout sheet specifying individual trials of the redesigned experiment to be performed by the user; h. performing an analysis of experimental data obtained by the user performing the trials of the redesigned experiment specified by the layout sheet created in step g.; and i. communicating a summary of the analysis performed in step h. to the user,

Claim 6

A computer implemented method of designing a robust experiment to be carried out by a user, the method comprising the steps of: a. inputting information to the computer, the information including: i. a response variable, ii. factors to be varied in the experiment, and iii. factor properties including indications representing which of the factors are noise factors and which of the factors are control factors; b. generating an experiment design in accordance with the information input to the computer in step a.; c. constructing two mathematical models, the first mathematical model containing terms representing hypothetical effects of all of the factors and the interactions between the factors on the response variable, the second mathematical model containing terms representing hypothetical effects of the control factors and interactions between control factors on the response variable; d. evaluating the experiment design generated in step b. by utilizing the information input in step a. and the two mathematical models constructed in step c. to compute values of statistical properties for terms in each of the two mathematical models; e. communicating the values computed in step d. to the user; f. generating a redesigned experiment by performing steps b. through e. using user changes to the information originally input in step a., until the user is satisfied with the values communicated in step e., wherein the user changes are suggested by a set of expert system forward chaining rules; g. creating a layout sheet specifying individual trials of the redesigned experiment to be performed by the user; h. performing an analysis of experimental data obtained by the user performing the trials of the redesigned experiment specified by the layout sheet created in step g.; and i. communicating a summary of the analysis performed in step h. to the user,

Claim 7

A computer implemented method of designing an experiment to be carried out by a user for studying complicated influence, the steps of the method comprising: a. inputting information to the computer, the information including: i. a response variable, ii. factors to be varied in the experiment, iii. factor properties including an indication of a number of factor levels for each of the factors, and iv. a number of times the experiment is to be repeated; b. generating a full factorial experiment design in accordance with the information input to the computer in step a.; c. constructing a mathematical model containing terms representing hypothetical effects of the factors and interactions between all of the factors on the response variable; d. evaluating the experiment design generated in step b. by utilizing the information input in step a. and the mathematical model constructed in step c. to compute values of statistical properties for terms in the mathematical model, including a minimum detectable difference value for each term in the mathematical model; e. communication the values computed in step d. to the user; f. generating a redesigned experiment of minimum size by repeating steps b. through e., while varying the number of factor levels of each of the factors originally input in step a. in accordance with a branch and bound search algorithm, until the minimum detectable difference values computed in step d. are less than user specified values; g. creating a layout sheet specifying individual trials of the redesigned experiment to be performed by the user; h. performing an analysis of experimental data obtained by the user performing the trials of the redesigned experiment specified by the layout sheet created in step g.; and i. communicating a summary of the analysis performed in step h. to the user.

Claim 8

A computer implemented method of designing an experiment to be carried out by a user for studying complicated influence, the steps of the method comprising: a. inputting information to the computer, the information including: i. a response variable, ii. factors to be varied in the experiment, iii. factor properties including an indication of a number of factor levels for each of the factors, and iv. a number of times the experiment is to be repeated; b. generating a full factorial experiment design in accordance with the information input to the computer in step a.; c. constructing a mathematical model containing terms representing hypothetical effects of the factors and interactions between all of the factors on the response variable; d. evaluating the experiment design generated in step b. by utilizing the information input in step a. and the mathematical model constructed in step c. to compute values of statistical properties for terms in the mathematical model, including a minimum detectable difference value for each term in the mathematical model; e. communication the values computed in step d. to the user; f. generating a redesigned experiment of minimum size by repeating steps b. through e., while varying the number of factor levels of each of the factors originally input in step a. in accordance with a branch and bound search algorithm, until the minimum detectable difference values computed in step d. are less than user specified values; g. creating a layout sheet specifying individual trials of the redesigned experiment to be performed by the user; h. performing an analysis of experimental data obtained by the user performing the trials of the redesigned experiment specified by the layout sheet created in step g.; and i. communicating a summary of the analysis performed in step h. to the user.