4 ms·

R isn't interactive or visual like Excel. Excel's learning curve is a lot less daunting. Also, Excel is ingrained in the accounting culture and well supported

R isn't interactive or visual like Excel. Excel's learning curve is a lot less daunting. Also, Excel is ingrained in the accounting culture and well supported by peers. Plus Excel is everyone's default database and form software. None of the article's reasons matter to non-IT people and no IT department has the juice to override the CFO and accountants. Unlike, every other interactive tool that displeases programmers, Excel survives. If you want R to replace Excel you need to build an interactive front end that can do everything Excel can as easily as Excel. Sitting in a class is not and option.

salamandersauce 5y ago
No, but it's far easier to see what the hell an R script is doing versus an Excel file. And there's just a ton more that R can do that Excel can't. More powerful plotting options, more tools like Random Forests or mixed effect models. And has less issues with floating point. Excel is good for accounting tasks because that's what it and other spreadsheet software were originally designed for. Data science tasks are tacked on.
azalemeth 5y ago
> Plus Excel is everyone's default database and form software. This is one of many, many things that is terribly wrong with Excel. It is arguably well suited for neither of those applications. I will continue to do my statistics in R, and be very glad that I do not work somewhere with either CFOs or powerful accountants ;-).
ineedasername 5y ago
Most of what you said only applies when the tasks are accounting related. There's a lot more to Excel, and a lot more than that to R. And just like, in accounting, you're not going to get the CFO to choose R over Excel, in many circles involving data analysis and statistics you won't get to choose anything but R. Also you don't need an interactive front end for R any more than you need an interactive front end for any other programming language. The scope of use is too large to allow for a single GUI. There are some packages that provide a GUI for specific purposes, like rattle, which provides an interface for common data mining tasks and models. I'd you want a pretty good spreadsheet GUI for viewing and modifying data, you can use rhandsontable. Just a few examples. Replacing Excel with R isn't necessary and I think would miss the point & relative strengths of the different tools. Recognizing where each tool is better suited for your needs, and if R is worth the learning curve for your specific needs, is what is important.
enumjorge 5y ago
Yeah most of the reasons explained in the article are too abstract for a beginner to understand. Visibility for example. If Excel is doing what the user needs, they might naturally question why they need to see under the hood. It might even be a positive that you get less visibility with it. Excel is exposing a smaller surface for the user to understand. Once you have a more complex task those guardrails are going to feel constraining, but until then Excel works well enough.

salamandersauce 5y ago

No, but it's far easier to see what the hell an R script is doing versus an Excel file. And there's just a ton more that R can do that Excel can't. More powerful plotting options, more tools like Random Forests or mixed effect models. And has less issues with floating point. Excel is good for accounting tasks because that's what it and other spreadsheet software were originally designed for. Data science tasks are tacked on.

azalemeth 5y ago

> Plus Excel is everyone's default database and form software. This is one of many, many things that is terribly wrong with Excel. It is arguably well suited for neither of those applications. I will continue to do my statistics in R, and be very glad that I do not work somewhere with either CFOs or powerful accountants ;-).

ineedasername 5y ago

Most of what you said only applies when the tasks are accounting related. There's a lot more to Excel, and a lot more than that to R. And just like, in accounting, you're not going to get the CFO to choose R over Excel, in many circles involving data analysis and statistics you won't get to choose anything but R. Also you don't need an interactive front end for R any more than you need an interactive front end for any other programming language. The scope of use is too large to allow for a single GUI. There are some packages that provide a GUI for specific purposes, like rattle, which provides an interface for common data mining tasks and models. I'd you want a pretty good spreadsheet GUI for viewing and modifying data, you can use rhandsontable. Just a few examples. Replacing Excel with R isn't necessary and I think would miss the point & relative strengths of the different tools. Recognizing where each tool is better suited for your needs, and if R is worth the learning curve for your specific needs, is what is important.

enumjorge 5y ago

Yeah most of the reasons explained in the article are too abstract for a beginner to understand. Visibility for example. If Excel is doing what the user needs, they might naturally question why they need to see under the hood. It might even be a positive that you get less visibility with it. Excel is exposing a smaller surface for the user to understand. Once you have a more complex task those guardrails are going to feel constraining, but until then Excel works well enough.