In monkeypicked/aa030pa: panel operations

Package aapa operates on the output from companion package aabd, applying more-or-less standard operations in a more-or-less standard sequence, with options to customise at each step. The underlying pattern uses function names in a numerical sequence, such that if all operations starting with a given number are complete, then operations on the next number can proceed. This defines processing 'chunks' which can be done in parallel, and 'stages' which need to be done serially.

Companion packages

• aautil : contains utility functions used in these examples and internally in the package.
• aabd : download is the package which generates the input panels, stored in RData
• aate : category builds and manages industry trees and fractional industry classification

Input and output data

• inputs are in root.global/BDH/derive-000, with one RData file per panel
• inputs go to root.global/BDH/derive-001, similarly structured

Session

In the following, it matters that aatopselect() is set before aapa is loaded, because the .onLoad sets global variables. [This is not totally satisfactory.]

require(aautil)
aatopselect("test")
require(aapa)
require(testthat)
deraapa()

After processing, access to the data is comparatively fast and scaleable

library(aapa)
aatopselect("test")
cor(getstep("x0703redv"),us='c')

Customisation

Adding additional functions to the package, with appropriate names, causes them to automatically be executed within deraapa() at a 'step' which corresponds to the numeric part of the name. So for example a function beginning 'x07' will be run at the end, and can take dependencies on all preceding steps.

monkeypicked/aa030pa documentation built on May 23, 2019, 6:10 a.m.