uq | Universal serialized data reader to JSON | JSON Processing library

What is the most efficient way to switch the locations of two columns in python datatable? I wrote the below function that does what I want, but this may not be the best way, especially if my actual table is big. Is it possible to do this in place? Am I missing something obvious?

I'm observing the following discrepancy in the efficiency of data.frame modification in R:

With the parent-child relationships data frame as below:

I have a vector with values (val) and a vector indicating group membership (group):

I have a dataset as follows -

I have the following toy data -

I have the toy data given below.

For a given positive integer n, I want to know the fastest base R (not Rcpp) algorithm for constructing the integer vector c(1:1, 1:2, ..., 1:n), which has length n*(n+1)/2. There are bonus points for fast and memory-efficient algorithms, since I ultimately want to avoid allocating a vector of length n*n.

I'm aware of at least two approaches:

• unlist(lapply(seq_len(n), seq_len), FALSE, FALSE)
• {J <- .row(c(n, n)); J[upper.tri(J, TRUE)]}

the latter being particularly inefficient since it allocates two integer vectors of length n*n.

Note that if we assign the value .col(c(n, n)) to J above, then we obtain the sequence 1 2 2 3 3 3 4 4 4 4 .... This sequence can be constructed fast and efficiently with {i <- seq_len(n); rep.int(i, i)}.

I am wondering if a similarly fast (or faster) algorithm exists in the .row(c(n, n)) case, or if unlist-lapply is optimal from a base R standpoint.

FWIW, here is a benchmark of the three procedures I've mentioned so far:

I am subtracting years from date column in Python which felt a lot slower than R which is usually not the case, so I am wondering is there a faster/ efficient way of doing this in python ?

(As after years of working in R I am moving back to python, so my python skills are not good anymore & looking for code optimization in python).