Avoid underscored function variants

R/data_preparation.R

@@ -23,7 +23,7 @@ auto_grouping <- function(data, input, target, n_groups, model="kmeans", seed=99
 
 	df_categ=categ_analysis(data, input , target)
 
-	d=select_(df_categ, "perc_target",  "perc_rows")
+	d=dplyr::select(df_categ, dplyr::all_of(c("perc_target",  "perc_rows")))
 
 	set.seed(seed)
 	if(model=="kmeans") {
@@ -44,12 +44,12 @@ auto_grouping <- function(data, input, target, n_groups, model="kmeans", seed=99
 	df_categ[, var_rec]=paste("group_", cluster_vec, sep = "")
 
 	## See new profiling based on new groups
-	data_rec=merge(select_(data, input, target), select_(df_categ, input, var_rec), by=input)
+	data_rec=merge(dplyr::select(data, dplyr::all_of(c(input, target))), dplyr::select(df_categ, dplyr::all_of(c(input, var_rec))), by=input)
 	recateg_results=categ_analysis(data_rec, var_rec, target)
 
 	l_res=list()
 	l_res$recateg_results=recateg_results
-	l_res$df_equivalence=arrange_(unique(select_(data_rec, input, var_rec)), var_rec)
+	l_res$df_equivalence=dplyr::arrange(unique(dplyr::select(data_rec, dplyr::all_of(c(input, var_rec)))), .data[[var_rec]])
 	l_res$fit_cluster=fit_cluster
 
 	return(l_res)

R/discretize.R

@@ -40,10 +40,10 @@ discretize_df <- function(data, data_bins, stringsAsFactors=T)
 
 	if(stringsAsFactors)
 	{
-		data_2b=data %>% mutate_at(vars(vars_num), conv_factor)
-		data_3=data_2b %>% mutate_at(vars(vars_num), funs(factor(replace(., is.na(.), "NA."))))
+		data_2b=data %>% dplyr::mutate(dplyr::across(dplyr::all_of(vars_num), conv_factor))
+		data_3=data_2b %>% dplyr::mutate(dplyr::across(dplyr::all_of(vars_num), ~factor(replace(., is.na(.), "NA."))))
 	} else {
-		data_3=data %>% mutate_at(vars(vars_num), funs(ifelse(is.na(.), "NA.", .)))
+		data_3=data %>% dplyr::mutate(dplyr::across(dplyr::all_of(vars_num), ~ifelse(is.na(.), "NA.", .)))
 	}
 
 	message(sprintf("Variables processed: %s", paste(vars_num, collapse = ", ")))
@@ -213,7 +213,7 @@ convert_df_to_categoric <- function(data, n_bins)
 	data_cat=discretize_df(data = data, data_bins = d_cuts, stringsAsFactors = F)
 
 	# Converting remaining variables
-	data_cat_2=data_cat %>% mutate_all(as.character)
+	data_cat_2=data_cat %>% dplyr::mutate(dplyr::across(dplyr::everything(), as.character))
 
 	return(data_cat_2)
 }

R/information_theory.R

@@ -23,7 +23,7 @@ entropy_2 <- function(input, target)
 
 	# get partial entropy
 	df_tbl=as.data.frame.matrix(tbl)
-	res_entropy=data.frame(t(df_tbl)) %>% mutate_all(funs(entropy(., unit = "log2"))) %>% head(.,1)
+	res_entropy=data.frame(t(df_tbl)) %>% dplyr::mutate(dplyr::across(dplyr::everything(), ~entropy(., unit = "log2"))) %>% head(.,1)
 
 	# computing total entropy
 	total_en=sum(probs_input*res_entropy)

R/models_lib.R

@@ -139,7 +139,7 @@ desc_groups <- function(data, group_var, group_func=mean, add_all_data_row=T)
 	stat=status(data)
 	vars_to_keep=stat[stat$type %in% c("integer", "numeric") & stat$variable != group_var, "variable"]
 
-	grp_mean=data %>% group_by_(group_var) %>% summarise_each_(funs(group_func), vars_to_keep) %>% mutate_each_(funs(round(.,2)), vars_to_keep)
+	grp_mean=data %>% dplyr::group_by(dplyr::across(dplyr::all_of(group_var))) %>% dplyr::summarise(dplyr::across(dplyr::all_of(vars_to_keep), group_func), .groups = "drop") %>% dplyr::mutate(dplyr::across(dplyr::all_of(vars_to_keep), ~round(., 2)))
 	grp_mean=data.frame(grp_mean)
 
 	grp_mean[,group_var]=as.character(grp_mean[,group_var])
@@ -149,7 +149,7 @@ desc_groups <- function(data, group_var, group_func=mean, add_all_data_row=T)
 
 	# vars_to_keep have all num variables (excluding group_var and factor/char). Calculate 'All_Data' means per column
 	data_num=select(data, one_of(vars_to_keep))
-	b=as.data.frame(data_num) %>% summarise_each(funs(group_func))
+	b=as.data.frame(data_num) %>% dplyr::summarise(dplyr::across(dplyr::everything(), group_func))
 
 	## putting all together: the sumarization per group plus the total per column
 	all_results=rbind(a, b)
@@ -194,7 +194,7 @@ desc_groups_rank <- function(data, group_var, group_func=mean)
 	vars_to_group=all_col[all_col!=group_var]
 
 	# mutate each does the group by only for variables defined in vars_to_group
-	d_group_rank=d_group %>% mutate_each_(funs(dense_rank(desc(.))), vars_to_group)
+	d_group_rank=d_group %>% dplyr::mutate(dplyr::across(dplyr::all_of(vars_to_group), ~dplyr::dense_rank(dplyr::desc(.))))
 
 	return(d_group_rank)
 }

R/target_profiling.R

@@ -71,7 +71,7 @@ get_target_plot <- function(data, input, target, plot_type)
 
 histdens_target <- function(data, input, target)
 {
-	cdf=group_by_(data, target) %>% summarise_(var.mean=interp(~mean(v, na.rm=T), v=as.name(input)))
+	cdf=data %>% dplyr::group_by(dplyr::across(dplyr::all_of(target))) %>% dplyr::summarise(var.mean = mean(.data[[input]], na.rm=TRUE), .groups = "drop")
 
 	cdf$var.mean=round(cdf$var.mean, 2)
 
@@ -167,13 +167,14 @@ categ_analysis_logic <- function(data, input, target)
 	tot_pos=sum(data[,target]==pred_class)
 
 	## profiling
-	grp=group_by_(data, input) %>% summarise_(
-					mean_target=interp(~round(mean(var==pred_class, na.rm = TRUE), 3), var = as.name(target)),
-					sum_target=interp(~sum(var==pred_class, na.rm = TRUE), var = as.name(target)),
-					perc_target=interp(~round(sum(var==pred_class, na.rm = TRUE)/tot_pos,3), var = as.name(target)),
-					q_rows=~n(),
-					perc_rows=~round(n()/nrow(data), 3)
-	) %>% arrange(-mean_target)
+	grp=data %>% dplyr::group_by(dplyr::across(dplyr::all_of(input))) %>% dplyr::summarise(
+					mean_target=round(mean(.data[[target]]==pred_class, na.rm = TRUE), 3),
+					sum_target=sum(.data[[target]]==pred_class, na.rm = TRUE),
+					perc_target=round(sum(.data[[target]]==pred_class, na.rm = TRUE)/tot_pos,3),
+					q_rows=dplyr::n(),
+					perc_rows=round(dplyr::n()/nrow(data), 3),
+					.groups = "drop"
+	) %>% dplyr::arrange(-mean_target)
 
 	#colnames(grp)[colnames(grp)=='sum_target']=paste("sum", target, sep="_")
 	#colnames(grp)[colnames(grp)=='perc_target']=paste("perc", target, sep="_")