Sum only numerical columns and divide values by total

Question

I am having trouble with some calculations on a data frame.

Here is my DF (with many more rows and columns)

What I am trying to do is:

Step (1) - For e

Answer 1

We can first separate "Assets" and "Returns" columns and then use colSums and multiply values accordingly

asset_col <- grep("^Assets", names(df1))
return_col <- grep("^Returns", names(df1))
colSums(t(t(df1[asset_col])/colSums(df1[asset_col])) * df1[return_col])

#Returns_Jan_2018 Returns_Feb_2018 
#        3.504230         4.633941 

---

To break it down and make clear each step

Step 1 - For each month I would like to sum the assets columns

colSums(df1[asset_col])
#Assets_Jan_2018 Assets_Feb_2018 
#    1466742         2049689 

Step 2 - For each firm, I would like to divide assets each month by the total for the month

t(t(df1[asset_col])/colSums(df1[asset_col]))
#     Assets_Jan_2018 Assets_Feb_2018
#[1,]      0.14333400      0.11485889
#[2,]      0.08395751      0.06202112
#[3,]      0.61217106      0.38532577
#[4,]      0.16053744      0.43779422

Step 3 - Then I would like to take the values from step (2) and multiply by the corresponding returns

t(t(df1[asset_col])/colSums(df1[asset_col])) * df1[return_col]

#  Returns_Jan_2018 Returns_Feb_2018
#1        0.6450030       0.76955455
#2        0.4449748       0.07442534
#3        0.8570395       2.38901980
#4        1.5572131       1.40094151

Step 4 - I would like to sum each column in step (3)

colSums(t(t(df1[asset_col])/colSums(df1[asset_col])) * df1[return_col])

#Returns_Jan_2018 Returns_Feb_2018 
#        3.504230         4.633941