AVM/RCA-apovshimben

Descript:

For AVM/RCA data clean, there are some function support query predict information from db and read .Rdata

Normally engineer will prefer to query predict_data from db and with this information to read those nenecessary .Rdata.

But this module support not only get those .Rdata from db's information but also can with the time interval to be the condition, this module can help you do some test on client envrionment to read .Rdata and no need db support, Good luck.

Usage:

First you shoud build a env.R file, this is a environment variable for DB and Rdata path. But if not, setting both variable psql_db_info and PATH are necessary.

# ex:
psql_db_info$psql.dbname <- ''
psql_db_info$psql.host <- ''
psql_db_info$psql.port <- {numeric}
psql_db_info$psql.username <- ''
psql_db_info$psql.password <- ''

PATH <- '.\\Batch_Output\\AVM_Model\\'


# or other ex:
psql_db_info <- list(psql.dbname='', psql.host='', psql.port={numeric}, psql.username='', psql.password='')

PATH <- '.\\{path}\\'

Then you can get rdata and predictx data like below:

# Loading .R file.
> source("avm.R")

# setting db config and Rdata path
psql_db_info <- list(psql.dbname='', psql.host='', psql.port={numeric}, psql.username='', psql.password='')
PATH <- .\\{path}\\''


# Get .Rdata with time interval, and return list data structure.
# get_traingingx_by_local(start.time, end.time)
> rdata <- get_trainingx_by_local('2017-12-05 00:00:00', '2017-12-05 14:00:00')

# Get .Rdata with db's information, and return list data structure.
# get_trainingx_by_db(psql_db_info, toolid, chamber, recipe, ystatistics, ysummary_value_hat_lower, ysummary_value_hat_upper, start.time, end.time)
# only  ystatistics, ysummary_value_hat_lower, ysummary_value_hat_upper is *float* type, other parametr is *str* type
> rdata <- get_trainingx_by_db(psql_db_info, 'CVDU01', 'P6|A5', 'UPAN120Q275A45|P-ANOA-A2-267X','l2tfin_uniform', 0, 0.1, '2017-09-21 23:00:00', '2017-09-24 03:00:00')

# Get predict x
# get_predictx(psql_db_info, toolid, chamber, recipe, ystatistics, ysummary_value_hat_lower, ysummary_value_hat_upper, start.time, end.time)
# only  ystatistics, ysummary_value_hat_lower, ysummary_value_hat_upper is *float* type, other parametr is *str* type
> predict.x <- get_predictx(psql_db_info, 'CVDU01', 'P6|A5', 'UPAN120Q275A45|P-ANOA-A2-267X','l2tfin_uniform', 0, 0.1, '2017-09-21 23:00:00', '2017-09-24 03:00:00')

But if you want to with glassid be condition, please follow below flow:

# Loading .R file.
> source("avm.R")

# setting db config and Rdata path
psql_db_info <- list(psql.dbname='', psql.host='', psql.port={numeric}, psql.username='', psql.password='')
PATH <- .\\{path}\\''


# Get .Rdata with db's information, and return list data structure.
# get_single_trainingx_by_db(psql_db_info, glassid, toolid, chamber, recipe, ystatistics)
# ALL parameters are *str* type
> rdata <- get_single_trainingx_by_db(psql_db_info, 'TL7CC0MAX', 'CVDU01', 'P2|A5', 'UPAN120Q275A45|P-ANOA-A2-267X', 'l2tfin_avg')

# Get single predict x
# get_single_predictx(psql_db_info, glassid, toolid, chamber, recipe, ystatistics)
# only glassid can support character, vector or list type other parameters are *str* type

#character
> single.predict.x <- get_single_predictx(psql_db_info, 'TL7CC0MAX', 'CVDU01', 'P2|A5', 'UPAN120Q275A45|P-ANOA-A2-267X', 'l2tfin_avg')
#list
> single.predict.x <- get_single_predictx(psql_db_info, list('TL7CC0MAX','TL79M07AF'), 'CVDU01', 'P2|A5', 'UPAN120Q275A45|P-ANOA-A2-267X', 'l2tfin_avg')
#vector
> single.predict.x <- get_single_predictx(psql_db_info, c('TL7CC0MAX'), 'CVDU01', 'P2|A5', 'UPAN120Q275A45|P-ANOA-A2-267X', 'l2tfin_avg')

RCA Function

This is root cause analysis scheme for AVM, below will descript how to combine with RCA-apovshimben. Which simply means, the RCA_Functiom.R need the data source get from RCA-apovshimben.

Usage:

With time interval and ysummary_value

# Loading .R file.
source("avm.R")
source("RCA_function.R")

# get rdata with DB, if return NULL mean no data
rdata <- get_trainingx_by_db(psql_db_info, toolid, chamber, recipe, ystatistics, ysummary_value_hat_lower, ysummary_value_hat_upper, start.time, end.time)

# get predict.x with DB, if return NULL mean no data
predict.x <- get_predictx(psql_db_info, toolid, chamber, recipe, ystatistics, ysummary_value_hat_lower, ysummary_value_hat_upper, start.time, end.time)

# execute the RCA_Function, handle by ma
ret <- RCA_func(rdata, predict.x)
return (ret)

With glassid

# Loading .R file.
source("avm.R")
source("RCA_function.R")

# get rdata with DB, if return NULL mean no data
rdata <- get_single_trainingx_by_db(psql_db_info, glassid, toolid, chamber, recipe, ystatistics)

# get single predict.x with DB, if return NULL mean no data
single.predict.x <- get_single_predictx(psql_db_info, glassid, toolid, chamber, recipe, ystatistics)

# execute the RCA_Function, handle by ma
ret <- RCA_func(rdata, predict.x)
return (ret)

or you can just call the api, the benefit is already do some basic error handling

# Loading .R file.
source("rca_api.R")

# with time and ysummary_value
rca <- get_rca(psql_db_info, toolid, chamber, recipe, ystatistics, ysummary_value_hat_lower, ysummary_value_hat_upper, start.time, end.time)

# with glassid
single.rca <- get_single_rca(psql_db_info, glassid, toolid, chamber, recipe, ystatistics)