Numbers - first thoughts

[dr_lha]

OK. I've been trying out the trail of Numbers. I'm a big Excel user at work so had some high hopes Apple could come up with something new and different.

My first impression is just how ordinary it is, I mean - it basically is a spreadsheet that does everything you'd expect a spreadsheet to do. Sure it has some Apple flash, and the page layout stuff looks nice, but other than that it seems a surprisingly complete spreadsheet program.

The layout is familiar if you're used to Keynote or Pages, a sparse toolbar and "Inspector" However there is also a 2nd toolbar with various options. The font size on this thing is tiny, pretty hard to read if you don't have great eyesight. Luckily I do, my wife doesn't and complained about it. The layout of the main toolbar is very customisable, the 2nd toolbar doesn't seem to be changeable or removable, unless I'm missing something.

However there are some issues. Firstly, the thing is as slow as molasses. OK I'll admit right now I'm trying it out on my 1Ghz G4 Powerbook. However I also use Excel 2004 on this machine and in comparision to Excel, Numbers crawls. Opening a Excel spreadsheet took minutes. OK, fair enough, conversion is a difficult thing. However I opened a CSV file, which is just a text file, and the conversion took about 90 seconds. The corresponding file opens in Excel in about 10 seconds on the same machine. I don't understand why a new app should be so sluggish. I remember people complaining about the same thing with Pages however. Do Apple write these apps in Applescript or something I'm going to install the Trial on my MacBook Pro tomorrow to see if its useable on a modern Mac, but honestly right now I couldn't recommend this to anyone who has a lower end G4 based Mac, whereas Microsoft Excel works great on those machines. OK its an older program, but it basically does the same thing and I see no reason why importing a CSV or scrolling through numbers should be juddery, its pretty basic stuff. I guess the flashy interface causes a high overhead on the program.

On the positive side, Excel importing did work surprisingly well, the figures looked the same as they did on Excel, slight smoother, but otherwise OK. A dialog box full of "warnings" popped up after the long wait to import. I couldn't be bothered to go through these, but they seem to be mainly incompatibilities. The file I imported was from Excel 2004, haven't tried any Open XML files yet (Office 2007 for Windows).

A couple of weird things: On a chart it seems like I can adjust the min/max and tick interval on one axis only. Odd, I'm sure I'm doing something wrong here. However the options for the charts seem minimal at best. Another thing is that you can't get a "+" symbol for the points if you're plotting an X/Y line plot, only a scatter plot! When sorting it didn't give me the option of having the top row as the name of the row. Excel has a checkbox.

Cool thing: When you select regions, statistics for the region you selected appear in the bottom left (mean, max etc).

Anyone care to share their thoughts so far. Right now I would consider myself both impressed (at the completeness) and frustrated (at the sluggishness).

 

[dr_lha]

Update. I just had to force quit the damn program because a relatively simple select action ("Show all rows where column K is not equal to this") basically lead to 10 mins to CPU 100%. Sigh.

 

[JaymanKC]

limits to utility

I fear that Numbers may be too limited for many scientific applications. I cannot find anywhere to input error bars. This is a killer for its utility in a lab setting.

Also, I have tried to open large Excel files of microarray data (45,000 rows and 6 columns), and Numbers chokes on it; I have to Force Quit.

Maybe I am missing something, but I do not think it will replace Excel (blah) in my lab. I really wanted to go Microsoft-free...

 

[nsbio]

I fear that Numbers may be too limited for many scientific applications. I cannot find anywhere to input error bars. This is a killer for its utility in a lab setting.

Also, I have tried to open large Excel files of microarray data (45,000 rows and 6 columns), and Numbers chokes on it; I have to Force Quit.

Maybe I am missing something, but I do not think it will replace Excel (blah) in my lab. I really wanted to go Microsoft-free...

Isn't Excel limited to some fairly low # of rows (60, 000), such that any whole-gemome array must be split into separate files? That seems to be a major bummer - I thought that iWork would be able to handle the extra load. Absence of error bars is another bummer - what were they thinking? One could always draw the error bars manually on the chart, but then one can always do a lab meeting with transparencies

 

[dr_lha]

I fear that Numbers may be too limited for many scientific applications. I cannot find anywhere to input error bars. This is a killer for its utility in a lab setting.

Ouch, I didn't notice that. Add that to my comment about the Charting options being limited. Error bars are a must for me too.

Also, I have tried to open large Excel files of microarray data (45,000 rows and 6 columns), and Numbers chokes on it; I have to Force Quit.

My experience opening a 700 row Excel file was bad enough. Importing stuff just seems to be slow.

Maybe I am missing something, but I do not think it will replace Excel (blah) in my lab. I really wanted to go Microsoft-free...

So far this is my conclusion. I'm waiting to see how it performs on my work machine though (the MacBook Pro). I do most of my plotting using other packages, so if it has some compelling features I might use it for some applications.

 

[CANEHDN]

From what I can tell Numbers is a great program. It's great for the average home user and does everything it needs to easily. It's uncluttered and sleek. I doubt it will have everything that Excel has but it's a great start. I love the ability to create multiple Sheets in the same window. Very Nice. I'm definitely buying this one.

The speed is extremely nice too. I've noticed no hiccups or pauses in anything I've tried so far.

 

[dr_lha]

The speed is extremely nice too. I've noticed no hiccups or pauses in anything I've tried so far.

OK. Good to hear its OK on an Intel Mac. My speed comments should clearly be taken in context with the fact I'm running on a old Mac. That said Excel works perfectly on my machine, and basically appears to do everything (and more) that Numbers does. What am I getting from Numbers for the extra CPU cycles it uses I wonder?

I just noticed another thing, I can't seem to change the size of the data markers on the plots, and by default they are huge and bold! Maybe I'm approaching this from too much of an Excel mindset, but charting seems to have some major deficiencies. This is supposed to be "iWork" though, not "iLikeToScrapBook", these apps are supposed to be useable for work, not just Home use.

 

[CANEHDN]

This is supposed to be "iWork" though, not "iLikeToScrapBook", these apps are supposed to be useable for work, not just Home use.

That's true. I bet once you dive more into the app, you'll find that a lot of the functions you expect are there. They've just given a different layout that we're not use to.

 

[macbwizard]

OK. Good to hear its OK on an Intel Mac. My speed comments should clearly be taken in context with the fact I'm running on a old Mac. That said Excel works perfectly on my machine, and basically appears to do everything (and more) that Numbers does. What am I getting from Numbers for the extra CPU cycles it uses I wonder?

I just noticed another thing, I can't seem to change the size of the data markers on the plots, and by default they are huge and bold! Maybe I'm approaching this from too much of an Excel mindset, but charting seems to have some major deficiencies. This is supposed to be "iWork" though, not "iLikeToScrapBook", these apps are supposed to be useable for work, not just Home use.

So far I'm loving it on my MBP CD (2gigs ram; 2.0 GHz). It's extremely fast on this machine, so I'd imagine any intel machine will run it great. I don't use that many advanced features so it works great for me. I also like the multiple charts in a single sheet.

 

[JaymanKC]

Isn't Excel limited to some fairly low # of rows (60, 000), such that any whole-gemome array must be split into separate files? That seems to be a major bummer - I thought that iWork would be able to handle the extra load. Absence of error bars is another bummer - what were they thinking? One could always draw the error bars manually on the chart, but then one can always do a lab meeting with transparencies

After processing with R, an Affymetrix dataset will be about 45,000 datapoints (for the mouse genome) per sample. Excel can handle it, but is not a speed demon. Numbers is choking even on the flat, tab-delimited file, with no calculations. I will try importing these files in Numbers on my lab's Mac Pro tomorrow, but I don't think the faster processor will help.

 

[Mac In School]

I'm not going to go as far as to say I'm not impressed, but I will say that early indications are that it won't work for me.

I work with data a lot more than I do actual spreadsheets, with calculations and stuff. I'd much rather use a database program for data (duh), but I have to work the files my client sends me, and they're almost always Excel files. It also makes sense for me to use a format I can import to MySQL via Navicat. So Excel files it is.

Keep in mind I did this on a pretty beefy 2-processor (4-core) machine with 4 GB of RAM, but I did run some tests:

Columns: 61
Rows 10,168

To open this as a CSV...

NeoOffice: 9 seconds
Numbers: 25 seconds

To open this as an Excel spreadsheet...

NeoOfffice: 3 seconds
Numbers: 18 seconds

To open the file after they were each saved to their native format:

NeoOffice: 5 seconds
Numbers: 6 seconds

Obviously it does okay with its own files, but is painfully slow to import. So, on to other criteria...

File format support: Unless I'm missing something Numbers seems to be extremely limited in this area. .CSV and Excel are fine. But no tab-delimited. Personally, I prefer to be able to choose any delimiter (I like the pipe |), just in case the data values contain commas.

My business parter does this fancy-ass form filtering stuff in Excel. I have no idea how he does it (and don't really care), but I need to be able to read and use it. I'm attaching a screen capture of what this looks like in Excel. Basically, the header cell is a dynamic list of all the values in that column. You choose any value, and it shows all the data that matches that value. This works fine in Excel and NeoOffice, but Numbers stripped it out.

I also had a few import errors in Numbers, where it didn't like the data it was being given. Is it possible that it does better error checking? Absolutely. But it works in Excel, NeoOffice, MS SQL, and MySQL, so chances are it's guessing what data should look like based on a few rows (and not all of them)... Therefore guessing wrong. MS Access does this all the @&*%ing time.

Looks like Numbers is probably a no-go for me, but I'm really liking what Iv'e seen in Pages so far.

 

[Anonymous Freak]

Well, I have a few old, convoluted Excel spreadsheets that have never worked in anything other than "Real" Microsoft Excel. So I downloaded the trial, and tried it on my monster, a 7.2 MB, 3 sheet spreadsheet where the main 'data' sheet has 1180 rows, 129 columns of data and formulas. (64 columns of data, 65 columns of formulas consisting mostly of 'lookup's on that row of data.)

A second sheet is a pure 'reference' sheet, of static data (mostly instructions on how to use the spreadsheet, what various acronyms throughout it mean, etc.)

A third sheet is the main 'lookup' sheet. It has an 8x44 section of extremely heavy calculation formulas for use in summarizing the data from the main data sheet. Lots of it is a 'static' set of summaries (the actual data isn't static, it's just that this section is 'view only',) and one section is where you can type in an acronym for what you are looking for, and it shows a summary of that. On my old PC notebook, a 2.26 GHz Pentium M with 1 GB of RAM, running Excel 2003, typing in a new acronym could take up to two minutes for it to calculate the new numbers. Numbers only takes about half a minute on my 2.0 GHz Core Duo, but I found out it's because it isn't doing all the work. This is what has never worked in other spreadsheet apps (Like OpenOffice.org.) The summary sheet relies on so many lookups, in some cases formulas that the individual formula is over 200 characters long. (Here's an example of one column's mess, yes, this is repeated 1180 times, in this column alone. Other columns are just as bad or worse: )

=IF(AND($P4=LookupSheet::Table 1 :: $A$39,$BO4),$O4*COUNTIF($BO4:$BV4,">0"),)+IF($T4=LookupSheet::Table 1 :: $A$39,$S4,)+IF($V4=LookupSheet::Table 1 :: $A$39,$U4,)+IF($Z4=LookupSheet::Table 1 :: $A$39,$Y4,)+IF($AB4=LookupSheet::Table 1 :: $A$39,$AA4,)+IF($AF4=LookupSheet::Table 1 :: $A$39,$AE4,)+IF($AH4=LookupSheet::Table 1 :: $A$39,$AG4,)+IF($AL4=LookupSheet::Table 1 :: $A$39,$AK4,)+IF($AN4=LookupSheet::Table 1 :: $A$39,$AM4,)+IF($AR4=LookupSheet::Table 1 :: $A$39,$AQ4,)+IF($AT4=LookupSheet::Table 1 :: $A$39,$AS4,)+IF($AX4=LookupSheet::Table 1 :: $A$39,$AW4,)+IF($AZ4=LookupSheet::Table 1 :: $A$39,$AY4,)+IF($BD4=LookupSheet::Table 1 :: $A$39,$BC4,)+IF($BF4=LookupSheet::Table 1 :: $A$39,$BE4,)+IF($BJ4=LookupSheet::Table 1 :: $A$39,$BI4,)+IF($BL4=LookupSheet::Table 1 :: $A$39,$BK4,)

Unfortunately, Numbers doesn't support 'ISERR' (?!?!?!) or 'INDIRECT', while it does warn me about it with a little blue flag on each affected cell (according to the error dialog on opening, there are 43,749 cells with unsupported functions,) it just replaced those cells formulas with the last calculated value, which means my lookup doesn't work right anymore, even though it *LOOKS* like it works at first glance. (The numbers change, but not including ALL of the numbers they should, since lots of my lookups had their lookup data replaced with static, incorrect for the new source, data.)

Trying on a second, much less insanely complex spreadsheet that likewise has some compatibility issues with OpenOffice (no INDIRECTs in this one, but apparently there are some ISERRs,) doesn't work right, either. Again, it *LOOKS* at first glance to be right, but if you change some data, all of the cells don't recalculate correctly, because some formulas have been replaced with "last calculated value"s. (This one only had that happen to 99 cells, not forty three thousand, this one might be fixable by hand by finding a way to rewrite the affected cells to avoid 'ISERR', although I remember when coding this spreadsheet that ISERR was the only way to accomplish what I wanted.)

Finally, my third, smallest, and least complex spreadsheet, doesn't open because it's too large!! (264 KB vs. #2's 1.8 MB, and #1's enormous 7.2 MB.)

The two that do open also suffer from formatting issues, but those aren't dealbreakers. The turning formulas into static data is.

 

[Anonymous Freak]

what were they thinking?

They were thinking this is for marketing guys who need pretty spreadsheets, not people who need serious spreadsheets. That's the only conclusion I can come up with for its glaring flaws in the face of its wonderful presentation abilities.

 

[QuarterSwede]

After processing with R, an Affymetrix dataset will be about 45,000 datapoints (for the mouse genome) per sample. Excel can handle it, but is not a speed demon. Numbers is choking even on the flat, tab-delimited file, with no calculations. I will try importing these files in Numbers on my lab's Mac Pro tomorrow, but I don't think the faster processor will help.

As much as I'd love Numbers to be better than Excel and work well for you guys, I'm pretty sure it's really for more normal and average uses like financial, charts, etc. NOT lab use where you need thousands of data points. I mean, really, who else needs that many?

 

[Anonymous Freak]

My business parter does this fancy-ass form filtering stuff in Excel. I have no idea how he does it (and don't really care), but I need to be able to read and use it. I'm attaching a screen capture of what this looks like in Excel. Basically, the header cell is a dynamic list of all the values in that column. You choose any value, and it shows all the data that matches that value. This works fine in Excel and NeoOffice, but Numbers stripped it out.

That is "AutoFilter". Select your batch of data, then go to the 'Data' menu, then the 'Filter' submenu, and choose "AutoFilter", it turns it into this dynamic filter for you. I haven't found an equivalent in Numbers, yet. (I use it a lot, too.)

 

[Mac In School]

I didn't want to know!!!



Dude, if he finds out I know how to do it, I'm screwed. I'll have to start doing it myself.

Thanks, though.

 

[balamw]

They were thinking this is for marketing guys who need pretty spreadsheets, not people who need serious spreadsheets. That's the only conclusion I can come up with for its glaring flaws in the face of its wonderful presentation abilities.

Sounds much like my reaction to the charting tool in every version of Excel I've used since the mid-80s, when I stopped using Cricket Graph and started doing many simple charts in Excel, at least Apple doesn't seem to be giving you tolls that will be barely adequate, but not good, at these tasks.

I've come to view it this way:

Spreadsheets (Excel) are like the swiss army knife of data analysis tools. Flexible and powerful enough to be dangerous, and good enough if you're only doing a task once or twice. For anything more serious, use a better suited tool be it a database, or a statistics/charting package, or write some code.

B

 

[JaymanKC]

As much as I'd love Numbers to be better than Excel and work well for you guys, I'm pretty sure it's really for more normal and average uses like financial, charts, etc. NOT lab use where you need thousands of data points. I mean, really, who else needs that many?

This is true, and even Excel is not really an appropriate means of analyzing this type of data. However, as a proof of principle, at least it is feasible to open and process these large files in Excel, and in Neoffice as well. Those applications are the baseline by which Numbers will be compared, and so far, there are some limits to this program compared to the competition.

The seeming lack of error bars in Numbers is what is really killing me. Even a high school student, which is an entirely appropriate target user, needs error bars for chemistry class. Has anyone found support for error bars in Numbers?

 

[QuarterSwede]

However, as a proof of principle, at least it is feasible to open and process these large files in Excel, and in Neoffice as well. Those applications are the baseline by which Numbers will be compared, and so far, there are some limits to this program compared to the competition.

I can understand your thought process but Apple has been known to not go directly against the competition and instead reinvent their apps to make them easier to use (and sometimes more limited) than the competition. While it's hard not to compare it to Excel you can't really compare it to NeoOffice or OpenOffice.org because the goal of those is emulate Microsoft Office as closely as possible (kind of redundant).

 

[Mac In School]

I can only speak for myself (and I realize you weren't addressing me), but I'm not necessarily trying to compare Numbers to Excel or NeoOffice. I'm merely testing it to see if I can get my job done better.

 

[balamw]

After playing with the trial for a bit, I must say that I really like the way it handles "tables" within a sheet. This is a much more "natural" workflow for me and would help organize spreadsheets with lots of smaller sets of data in them. It also lends itself to easier formatting for printing for the kinds of things they show in the video tour.

That said, I HATE the charting tool for most of my needs. Why can't Apple or Microsoft build a decent scatter plot tool. Is it that hard?

Two other deficiencies I noted in it, besides the lack of error bars.

1. No easy way to plot multiple series without repeating the X axis, Say I want to plot X vs. Y and X vs. Z I need four columns X,Y,X,Z or to do the plot in two steps. [I need this all the time, often plotting 8 quantities vs one x axis. Easy in Excel.]
2. The legend took the header form the X axis data, so I ended up with two X entries in the legend.

B

 

[gauchogolfer]

So it seems from the comments here that I need to keep waiting for Origin to come out with an OS X version...fat chance.

 

[balamw]

So it seems from the comments here that I need to keep waiting for Origin to come out with an OS X version...fat chance.

Parallels/VMWare are your friend.

EDIT: Have to try this when I'm done with messing around with Numbers. http://plot.micw.eu/

B

 

[gauchogolfer]

Parallels/VMWare are your friend.


B

Not on my powerbook, they aren't....

 

[Nermal]

I must going blind; where do you download the trial from?

Edit: Found it on the US site. It's not on the NZ one.