R/Subset.R

In tohweizhong/redhyteR: A hypothesis mining package in R

# Subset.R (internal)

Subset <- function(df, Atgt, Acmp, Atgt_cl = "", Acmp_cl = "", Actx_items = "", ht = NULL){

    require(magrittr)
    set.seed(123)

    # Atgt
    if(ht[["Atgt_type"]] == "num"){
        idx_tgt <- 1:nrow(df)
    }
    else if(Atgt_cl %>% length > 1){
        idx_tgt <- NULL
        for(cl in Atgt_cl){
            idx_tgt <- c(idx_tgt, which(df[,Atgt] == cl))
        }
    }

    # Acmp - must be categorical
    if(Acmp_cl %>% length > 1){
        idx_cmp <- NULL
        for(cl in Acmp_cl){
            idx_cmp <- c(idx_cmp, which(df[,Acmp] == cl))
        }
    }

    # take intersect
    rows   <- intersect(idx_tgt, idx_cmp)
    df_ctx <- df[rows,]
    df_ctx <- droplevels(df_ctx)

    # now for Actx
    # Actx_items is the format of
    # c("A1 = c1", "A2 = c2", ...)
    if(Actx_items %>% nchar > 0){
        A       <- NULL
        cl      <- NULL
        idx_ctx <- df_ctx %>% nrow %>% seq
        for(ii in Actx_items){
            tmp  <- strsplit(ii, " = ") %>% unlist
            A    <- c(A,  tmp[1])
            cl   <- c(cl, tmp[2])
            print(A); print(cl)

            tmp_idx <- which(df_ctx[,A] == cl)
            idx_ctx <- intersect(idx_ctx, tmp_idx)
        }
        df_ctx <- df_ctx[idx_ctx,]
    }

    return(df_ctx)
}