add examples/exercises

[franknarf1]

(looking at those i contributed to Docs.SO...)

Cleaning data

• Handling duplicates: DT = data.table(id = c(1,2,2,3,3,3))[, v := LETTERS[.I]][]
  • keep one row per group
  • keep only unique rows
  • keep only nonunique rows
• unlist list column https://stackoverflow.com/questions/38796767/set-operation-within-a-list-column
• stack related cols with melt's patterns()
• dcast census data (with multiple metrics arrayed on rows)

... going beyond Docs.SO:

• nice explanation of the advantages of long data for analysis vs wide data for printing out https://gpavolini.netlify.com/2019/07/22/long-to-wide-and-wide-to-long-format-conversion and https://twitter.com/gpavolini/status/1153466256542113792

---

Creating a data.table

A data.table is an enhanced version of the data.frame class from base R. As such, its class() attribute is the vector "data.table" "data.frame" and functions that work on a data.frame will also work with a data.table. There are many ways to create, load or coerce to a data.table, as seen here.

• modify data.frame by reference
• rbindlist a bunch of sub-tables built with data.table()
• coerce a data.frame
• copy()
• fread()

---

Adding and modifying columns

... hm, this is getting somewhat tedious. i will dump the contents into a new file, exercises.something

Another: https://stackoverflow.com/q/49282735/

[franknarf1]

Possible typos in existing examples:

• library(package = "_your_new_package_"), should be help(package = "_your_new_package_")?

Misc additions

• Re basics, just link to Q&A re loop traps https://stackoverflow.com/q/52597296 though really it applies beyond R, with any mutable objects

• Re sets (under "getting work done"), anyDuplicated ("Read the docs, use it with big table x, note the speedup vs any(duplicated(x))")

• Similarly, re missing values (under "basics"), anyNA

• In Basics > Vectors, the gl() function for making fancy sequences (I'm also thinking about use of grouping() which seems to supply data.table internals...)

• shifting a single column by group where fill= is a constant can be addressed by sorting on grouping vars and using ungrouped shift. Also applies to x - shift(x, fill=first(x)) or x > shift(x, fill=first(x)), since those are 0 & FALSE by construction https://github.com/Rdatatable/data.table/issues/1534 (just saved myself from a long computation by searching the issues for this...)