Poll of the Day > How do you feel about excel

(+) Yeah! (+)

I hate excel. It thinks it knows better than you, the user, and automatically converts types. There's a bunch of unseen metadata. It's horrible for large data sets.

Give me a plaintext tab-delimited file that I can process in R, please and thank you

---

Good for normal day-to-day calculations. Personally, I like using it as a makeshift calculator, but I imagine why it would be unwieldy for people pursuing statistical work.
---

I think its convenient and can do a lot of cool things. And I do think its better than Google Sheets.

But everything Clench said is right.

---

EclairReturns posted...

Good for normal day-to-day calculations. Personally, I like using it as a makeshift calculator, but I imagine why it would be unwieldy for people pursuing statistical work.

This is basically what I use it for. And things like tracking budgets. Small things.

---

Clench281 posted...

I hate excel. It thinks it knows better than you, the user, and automatically converts types. There's a bunch of unseen metadata. It's horrible for large data sets.

Give me a plaintext tab-delimited file that I can process in R, please and thank you

aka eve online?

I don't hold excel on any high pedestal I only use it for the simple things it can do & thats it.

---

It's a pain to use but very powerful.

---

It's very good for organizing and presenting data in a user-friendly form, and tremendously versatile once the user knows what to google to figure out the best functions for their desired purpose, but I can see that it would fall short for more advanced statistical analyses.

---

It peaked in 2003, but then M$ kept releasing new editions for $$$. It used to be about the artistry, M$. What happened to that?!

In the office, I'll use Excel but at home I use Apache OpenOffice.

---

Its vastly more feature rich and powerful than google sheets.

It's also broken and useless at my work because of their shitty implementation of a shared drive, making data sharing between multiple spreadsheets completely fubar.

So google sheets it is.

---

Zeus posted...

It peaked in 2003, but then M$ kept releasing new editions for $$$. It used to be about the artistry, M$. What happened to that?!

I used '07 quite happily for many years, but it's only in '10+ that you get access to pivot tables, which are really handy for some specific applications. I think it's even later than that that they introduced the ability to incorporate cell arrays into formulas, which is pretty niche, but when it's what you need, it's a godsend. As I've delved deeper into Excel, I've come across more and more solutions that rely on tools that are only present in later versions, such that I don't really resent the subsequent releases.

The current subscription model can die in a fire, though. I get that they want an update-based model instead of releasing new versions every few years, but I'd rather pay once and be out of date for a couple iterations (see: using Office '07 pretty much right up until I replaced my desktop's SSD last year and lost the installed copies) than pay every year for them to pretend Word and Powerpoint have changed appreciably in 20 years (not entirely fair, Word 2003 was leagues better than 2000, but I digress).

---

Better to be excel than incel...

---

Clench281 posted...

It thinks it knows better than you, the user, and automatically converts types.

this is annoying, yes. you basically need to write various read functions that can each accept any kind of cell value, since a text column can have a number in it or vice-versa.

but then, I don't think there's that many options either. any cell is going to have either a string, a double, or null. (could be wrong, but I think double is the only numeric type, which is also used for date values.) so really your main difference is between text and numbers, and then just deciding how picky you want to be about truncating, or if a value is required, etc.

so, a read int function that truncates but requires a value (pseudo-code):

case null: throw "value is required in {row, col}"
case string: parse double, throw if cant, cast to int
case double: cast to int
case other: throw "unexpected data type in {row, col}"

you can have a complete collection of such functions, and then provide a mapping between column header values and read functions and you can load data independent of column order and report when a required column is missing.

also, it's a *lot* faster to load the whole sheet into an array rather than scan the sheet itself. just be aware that it will be an array with a 1-based index. (you can use the UsedRange property of the Worksheet object to get the entire range of data in the file (though in certain cases it might be way too big if someone placed a random value way out in the middle of nowhere)).

---

adjl posted...

I used '07 quite happily for many years, but it's only in '10+ that you get access to pivot tables, which are really handy for some specific applications.

Pivot tables were available in Excel since the 90s. >_>

And much of the time I'd rather use Access than pivot tables. Among other things, it tends to be a bit less quirky than pivot tables when it comes to handling data sets.

adjl posted...

The current subscription model can die in a fire, though. I get that they want an update-based model instead of releasing new versions every few years, but I'd rather pay once and be out of date for a couple iterations (see: using Office '07 pretty much right up until I replaced my desktop's SSD last year and lost the installed copies) than pay every year for them to pretend Word and Powerpoint have changed appreciably in 20 years (not entirely fair, Word 2003 was leagues better than 2000, but I digress).

Supposedly that's what M$ wanted to do all along, but they didn't really have the technology to make that sort of a thing feasible (in that they could monetize it effectively)

---

Zeus posted...

Pivot tables were available in Excel since the 90s. >_>

It may be that later versions updated their functionality somehow. All I know is that I've run into a couple of pivot table-based solutions to problems I was having that weren't possible to do in 07 (or at least, not without VBA or other such complexities).

Zeus posted...

And much of the time I'd rather use Access than pivot tables. Among other things, it tends to be a bit less quirky than pivot tables when it comes to handling data sets.

The few times I've used them, they've been a bit fiddly, but they've done the job well enough. That said, I've generally only used them for relatively small data sets (large enough that manually performing the desired analysis would be impractical, but not so large that a proper database would be reasonable). I can see that a proper database would be more sensible for larger-scale work.

Zeus posted...

Supposedly that's what M$ wanted to do all along, but they didn't really have the technology to make that sort of a thing feasible (in that they could monetize it effectively)

Wouldn't surprise me in the slightest. I still don't like it, though. Office is a case where I want a product, not a service.

---

As a former accounting student, I found it essewntial.

---

Excel isn't very good gum, they have this weird chemical in them that I can only describe as Ubermenthe...

---