bigdata - R bigglm ffdf returns size-related error -
i trying run logistic regression model on ffdf
using bigglm.ffdf
ffbase error:
x1 = as.ff(sample(0:5, 400, replace=t)) x2 = as.ff(sample(c(rep(0, 98), rep(1, 2)), 400, replace=t) ) df2 = ffdf(x1, x2) mod = bigglm.ffdf(x2 ~ x1, family = binomial(), data = df2) > error in r[ok, ok] <- chol2inv(r[ok, ok, drop = false]) : number of items replace not multiple of replacement length
any insight per wrong code? many in advance.
edit: ram objects run expected:
x5 = sample( 0:5, 400, replace=t) x6 = sample(c(rep(0, 98), rep(1, 2)), 400, replace=t) df3 = data.frame(x5, x6) mod = glm(x6 ~ x5, family = binomial(), data = df3) # no error
Comments
Post a Comment