Improve R speed when using dynamic column names - r

Question

I have some code that works as expected but is very inefficient and slow. I have 2 dataframes - 1) with 40,645 rows and 264 columns where each column represents a KPI/dimension of some kind and 2) with 478,872 rows and 11 columns. DF1 is a wide dataframe and DF2 is a long data frame. I need to merge the two but cannot simply merge the columns since the dataframes are different formats. Additionally I need to merge 2 columns values from the DF2 to create a name for the new column in DF1. In the end I am accomplishing this using a loop and this command to do the work is what is slowing the code down substantially:
DF1[[col_name_v]][index_v] <- col_val_v
# Additionally these other methods appears just as slow:
DF1[index_v,"col1"] <- col_val_v
DF1[index_v,265] <- col_val_v
If I actually where to specify each of the column names manually like this it works 10X+ faster:
DF1$col1[index_v] <- col_val1_v
DF1$col2[index_v] <- col_val2_v
DF1$col3[index_v] <- col_val3_v
#.... etc
The problem is I need the code to be dynamic because there are many columns and potentially those column names may change over time so I would prefer the code to dynamically learn the column names and apply them on the fly to prevent frequent additions and changes to the code.
Data looks like the following:
DF1 (40645*264) - before adding new columns from the code below
Date_Time,YEAR,MON,DOW,WEEK,Location,ID,KPI1,KPI2,KPI3,...,KPI264
9/7/2020,2020,September,Monday,36,33,33001,43,0,2,...,10
DF2 (478872*11) - multiple rows to be merged as multiple columns to DF1
Date_Time,Location,ID,Technology,Cluster,Dimension,Variable,Formula,Value,Index
9/7/2020,33,33001,1,"OVERALL","NEWKPINAME1","SUM(N)/SUM(D)",2.8,0.003
9/7/2020,33,33001,1,"LOCATION","NEWKPINAME1","SUM(N)/SUM(D)",2.8,0.004
9/7/2020,33,33001,1,"GROUP1","NEWKPINAME1","SUM(N)/SUM(D)",2.8,0.002
dimension+variable are combined to create a new unique KPI name to be added as a column and the index value is written to DF1 for that new column index location.
# Provides the number of new columns that need to be added to DF1
dim_v <- unique(DF2$Dimension)
var_v <- unique(DF2$Variable)
limit_v <- length(dim_v) * length(var_v)
# This part adds the new columns to the DF1 - with NA set for values
index_v <- 1
while(index_v <= limit_v) {
# Create the column name
col_name_v <- paste(DF2$Dimension[index_v], DF2$Variable[index_v],sep="_")
# Add the column name with default values of NA
DF1[[col_name_v]] <- NA
index_v <- index_v + 1
}
# This part writes the values of each DF2 Dimension_Variable KPI values stored in ROWS (12 per ID) to DF1 across 12 COLUMNS
# Merge the long DF2 ROW based KPIs into the wide DF1 COLUMN based KPIs
index_v <- 1
while(index_v <= nrow(DF1)) {
# Identify the key fields we need for the lookup
datetime_v <- as.Date(DF1$Date_Time[index_v])
id_v <- DF1$ID[index_v]
# Create a temp data.frame for the related data
data_df <- subset(DF2, Date_Time == datetime_v & ID == id_v)
# Do we even have records?
if (nrow(data_df) !=0) {
# cycle through and write each value
index2_v <- 1
while(index2_v <= nrow(data_df)) {
# Create the column name
col_name_v <- paste(data_df$Dimension[index2_v], data_df$Variable[index2_v],sep="_")
col_val_v <- data_df$Index[index2_v]
# Write the values related to the column name
DF1[[col_name_v]][index_v] <- col_val_v
index2_v <- index2_v + 1
}
} else {
print(sprintf("No records for Date: %s ID: %s", datetime_v, id_v))
}
index_v <- index_v + 1
}