glm: Logistic and Poisson Regression.

In Fuzzy-Logix/AdapteR: Converts R syntax to SQL to provide fast analytics using in-database C++ library DB Lytix.

Description

glm performs logistic and poisson regression on FLTable objects.

Usage

glm(formula, data = list(), ...)

Arguments

formula	A symbolic description of model to be fitted
data	An object of class FLTable
family	Can be one of poisson,binomial,logisticwt or multinomial characters. Can be family functions like stats::poisson wherever possible.
catToDummy	Transform categorical variables to numerical values either using dummy variables or by using Empirical Logit. If the value is 1, transformation is done using dummy variables, else if the value is 0, transformation is done using Empirical Logit.
performNorm	0/1 indicating whether to perform standardization of data.
performVarReduc	0/1. If the value is 1, the stored procedure eliminates variables based on standard deviation and correlation.
makeDataSparse	If 0,Retains zeroes and NULL values from the input table. If 1, Removes zeroes and NULL. If 2,Removes zeroes but retains NULL values.
minStdDev	Minimum acceptable standard deviation for elimination of variables. Any variable that has a standard deviation below this threshold is eliminated. This parameter is only consequential if the parameter PerformVarReduc = 1. Must be >0.
maxCorrel	Maximum acceptable absolute correlation between a pair of columns for eliminating variables. If the absolute value of the correlation exceeds this threshold, one of the columns is not transformed. Again, this parameter is only consequential if the parameter PerformVarReduc = 1. Must be >0 and <=1.
classSpec	list describing the categorical dummy variables.
whereconditions	takes the where_clause as a string.
pThreshold	The threshold for False positive value that a user can specify to calculate the false positives and false negatives. Must be between 0 and 1.
pRefLevel	Reference value for dependent variable in case of multinomial family.
maxiter	maximum number of iterations.

Value

glm returns FLLogRegrMN object for multinomial family and FLLogRegr otherwise

Constraints

The anova method is not yet available for FLLogRegr. In case of multinomial family, residuals,fitted.values properties are not available.plot,influence methods are also not available. Properties like print(fit$x),model,plot might take time as they have to fetch data

See Also

glm for R reference implementation.

Examples

deeptable <- FLTable(getTestTableName("tblLogRegr"),"ObsID","VarID","Num_Val",
               whereconditions="ObsID<7001")
glmfit <- glm(NULL,data=deeptable)
coef(glmfit)
summary(glmfit)
head(residuals(glmfit))
plot(glmfit)
glmfit <- glm(NULL,data=deeptable,family="logisticwt",eventweight=0.8,noneventweight=1)
summary(glmfit)
plot(glmfit)
connection <- flConnect(odbcSource = "Gandalf",database = "FL_DEV")
widetable  <- FLTable("siemenswidetoday1", "ObsID")
poissonfit <- glm(event ~ meanTemp, family=poisson, data=widetable,offset="age")
summary(poissonfit)
plot(poissonfit)
predData <- FLTable(getTestTableName("preddata1"),"ObsID")
mu <- predict(poissonfit,newdata=predData)
deeptable <- FLTable(getTestTableName("tblLogRegrMN10000"),"ObsID","VarID","Num_Val",
             whereconditions="ObsID<7001")
glmfit <- glm(NULL,data=deeptable,family="multinomial")
glmfit$coefficients
glmfit$FLLogRegrStats
glmfit$FLCoeffStdErr
summary(glmfit)
print(glmfit)

Fuzzy-Logix/AdapteR documentation built on May 6, 2019, 5:07 p.m.