Outstanding blog post which led me here. Particularly valued your code example comparing "old school" and the Tidyverse on the same task, and the following analysis of each.
It's a pity data.table is left out of this talk. It sorted out many of the problems mentioned and in a way has a deep relation to the "tidyverse", considering melt, dcast functions originally from reshape2 are in there.
data.table was one of the biggest advances R ever saw. TidyVerse is a new language and they push it a lot and try to ignore the rest or take it over as theirs.
learn Base-R first. Do not engage with TidyVerse until you absolutely have to. Otherwise, you will unable to document and debug your code. Then add data.table and packages that are relateable to base-R. And, though caret is fantastic, it is likely better to use the closer-to-base packages first and then learn caret. Otherwise you know too little about what is going on under the hood.
Outstanding blog post which led me here. Particularly valued your code example comparing "old school" and the Tidyverse on the same task, and the following analysis of each.
Great talk. It improved my awareness and understanding of R as a organic, developing system. Thanks!
Wickedly elegant answer about why R seems like a sloppy mess sometimes around 49:00 or so - Roger Peng in the best.
22:50 where one realizes, TidyVerse will take away your freedom but also give you some tools to easily create some standard solutions.
It's a pity data.table is left out of this talk. It sorted out many of the problems mentioned and in a way has a deep relation to the "tidyverse", considering melt, dcast functions originally from reshape2 are in there.
data.table was one of the biggest advances R ever saw. TidyVerse is a new language and they push it a lot and try to ignore the rest or take it over as theirs.
I was promised some tapply, but all we got was aggregate.
learn Base-R first. Do not engage with TidyVerse until you absolutely have to. Otherwise, you will unable to document and debug your code. Then add data.table and packages that are relateable to base-R. And, though caret is fantastic, it is likely better to use the closer-to-base packages first and then learn caret. Otherwise you know too little about what is going on under the hood.
Even from the very beginning, R has been only superficially similar to S, just as Julia is superficially similar to MATLAB.