"Why do we use R rather than Excel?"

@edent — Sun, 11 Jul 2021 11:08:15 +0000

I recently had cause to take a beginners course in R - a language I'm fairly familiar with.

One of the other students had never used it before, so we were buddied up in order for me to show them the ropes.

The first lesson of R is always the same. Read a CSV, manipulate it a bit, draw a graph. We did it all without much fuss - and a graph appeared on screen. Nifty!

"I don't get it," the student said, "Why wouldn't you just use Excel for this?"

To a programmer, it seems obvious - but it's a fair question! If you have a static set of data, you can drag your mouse over it, hit a few buttons, and a graph appears. Much easier than wrestling with esoteric syntax in a text-based interface. This isn't a Koan - where the student becomes enlightened at the end. It's a tricky question to answer. Here are the reasons I gave - feel free to add your own.

Visibility

How do you see the code inside an Excel document? How do you tell exactly what is going on? You have to go clicking through cells, or reverse engineer what settings a graph has.

With something like R, you automatically have all the code visible in front of you. Reading through the code in a linear fashion is possible. You can trace exactly what the code is doing without having to worry about whether there's some code hidden in cell Z44.

Track Changes

Related to the above, it's hard to visualise what changes have been made to an Excel document. I don't know any way to easily see how a formula has changed. With R and Git (or any other version control system) you can see exactly what has changed from one version to the next.

Repeatability

Typically, a user draws a graph on a single Excel document. If you want the same graph of a different data set, you're out of luck. You can copy the data from one Excel sheet into another - but there's no way to easily copy a bunch of manipulations and graph configurations to another document.

With R, you just change read.csv("1.csv") to read.csv("2.csv") and the exact same calculations are run on two different data sets.

Batch processing

Related to the above, you can read every CSV in a directory and produce a graph for each of them. You can read data from an API and run the same process on it that you did yesterday.

Extensibility

Excel has a wide range of graphs available - but R has more. Excel can do basic analysis - but R can do extensive, complex analysis. Excel has some decent tooling - but R has thousands of libraries which can do a bewildering array of clever stuff.

What else?

Those were the advantages that I could think of on the spur of the moment. Perhaps you can think of more.

But what I learned was that it is decidedly non-obvious why a user would want to use something like R or Python when Excel (seemingly) covers all the basics.

Excel as a mapping tool

@edent — Fri, 23 Apr 2021 11:23:19 +0000

About a million years ago, my wife's work sent her on a training course optimistically titled "How To Use Excel As A Database."

We were both horrified. Excel is a perfectly good spreadsheet program - but it is categorically not a database!

OK, it has rows and columns which sorta look like a database table. And you can put constraints on cells which mimic a schema. And, yes, you can sort and query data. And you can join data across multiple tables. And... you know what...? Excel is a pretty decent visual introduction to databases. Sure, it isn't ACID and you wouldn't want to run a production environment off it. But Excel is a reasonable introduction to some database concepts.

My wife had to visit a hospital recently for a COVID test. Hospitals are often large and confusing buildings. But the hospital's website had a handy "Click here for a map" link. She clicked it, and I heard her scream from across the house.

What?

WHAT?!!?

Yes, some enlightened person had decided to use Excel to draw a building map... I can't even...

There are many reasons why this is bad. A user might not have Excel installed on their phone. The thin grey lines are easy to confuse with thick grey lines. There's no way to see entrances. The plain text means it doesn't have familiar icons for toilets. I'm sure you can think of a dozen more reasons.

And yet... It was probably easier to get permission to install Excel on a machine than it was to procure a specialist graphics tool. Just about anyone can use Excel to draw boxes without training. It's easy to edit if a room is closed or changes function. The text is searchable, unlike a graphic. It doesn't overwhelm the user with colours and extraneous information. At 41KB it's easy to download over a crappy WiFi connection.

As a quick hack, to get a vaguely usable map up on the website, Excel is... well, it isn't fine. It is barely even adequate. But it fulfils the brief. I hope that this was a quick bodge done one day with the aim of fixing it later. But, 7 years later, it is still there. Do people complain, or do they sigh and move on?

Or am I wrong and Excel is the perfect tool for the job?

You can see people's reactions to this map in the following Twitter thread.

excel – Terence Eden’s Blog