In decisionpatterns/stitch: stitch - Safe Left Joins for Data Rectangularizaton

These are the rational for this package:

data.table provides an inconvenient LEFT JOIN mechanism.

In data.table the LEFT JOINS occur by putting the LHS of the join on the right, e.g.

 X[DT, on="x"]                         # left join

where DT is the LHS join table. This just looks ugly.

COLUMN NAME COLLISIONS are not insufficiently disambiguated

data.table prefixes column-name collisions with x. and i.. dplyr suffixes with .x and .y. These names are not descriptive enough for joins where lots of tables are brought together.

COLUMN ORDERING is not (always) preserved.

The column ordering should be preserved keeping LHS columns on the LHS.

decisionpatterns/stitch documentation built on Sept. 9, 2020, 12:43 a.m.