Slashdot Mirror
Advanced Excel for Scientific Data Analysis
cgjherr writes "If the recent financial meltdown has left you wondering, 'When does exponential decay function stop?' then I have the book for you. Advanced Excel for Scientific Data Analysis is the kind of book that only comes along every twenty years. A tome so densely packed with scientific and mathematical formulas that it almost dares you to try and understand it all. A "For Dummies" book starts with a gentle introduction to the technology. This is more like a "for Mentats" book. It assumes that you know Excel very well. The first chapter alone will have you in awe as you see the author turn the lowly Excel into something that rivals Mathematica using VBA, brains, and a heaping helping of fortitude." Read on for the rest of Jack's review.
Advanced Excel for Scientific Data Analysis
author
Robert de Levie
pages
700
publisher
Oxford Press
rating
9
reviewer
Jack Herrington
ISBN
9780195370225
summary
Use Excel for high end scientific data analysis akin to Mathemetica
When I first opened this book my mouth just dropped. It had been years since I had seen a book typeset using LaTeX. But in an instant it made sense as the book is crammed packed with the kind of equations that would have been a nightmare to build with any other tools. Chapter after chapter has everything a really smart person needs to do curve fitting, statistical measures, differential equations, time-frequency analysis. But don't expect a play by play here. You will get the equations, set within a few dense paragraphs, with maybe a spreadsheet and a chart or two to show the results.
The first chapter concentrates on the getting the most out of Excel as a tool. All the chapters that follow dig into specific data analysis techniques. Chapters two, three and four are on least squares. Chapter five and six cover the analysis in the time domain including fourier transforms. Chapter seven covers differential equations. Chapter eight returns to Excel by digging in deeper into macros. Which leads into chapter nine, where we dig deeper into basic mathematical operations. Chapter ten covers matrix operations. And chapter eleven wraps it all up by giving you some spreadsheet best practices.
In University style there are also some exercises that you can do along the way if you want to tweak your brain pan a little more. To amuse myself I tried a few and I believe the book would have assessed my attempts 'wanting' if it had a voice to tell me.
Where most books like this would have several authors this book has just one; Roberte de Levie. This means that the tone, style and quality of the book is consistent throughout. A fact that you will come to appreciate as the book wades in ever increasingly deep data analysis concepts as the chapters roll on.
Though I would have preferred the book to have code samples in C#, I understand that the language of Excel is VBA and I guess I have to live with that. Thankfully VBA has come a long way and if you so inclined it would likely be easy to translate the code into C#, Java, or whatever else you like.
The fact that one person wrote the book left me wondering, "Who is this guy?" In my minds eye I kinda of figured he would look like one of those pulsing brain guys from Star Trek. Turns out he is a professor at Bowdoin College. And his fields of study include ionic equilibria, electrochemical kinetics, electrochemical oscillators, stochastic processes, and a whole lot more stuff that almost seems made up to sound impressive.
When this book isn't serving as an amazing reference for both Excel, scientific problem solving, or just insane equations it serves other purposes as well. It's a handy portable IQ test, as the count of pages you can grind through in one sitting, plus 90, is roughly your intelligence quotient. And if you fail at that you can always put a copy of the book, along with the Orange Bible, under your pillow and try to osmose your way to becoming the Kwisatz Haderach.
In all seriousness, this is a great book. It represents the kind of in-depth work and research we used to see in books that came out twenty years ago. Robert is to be applauded for his work. This is an excellent resource for anyone looking to do scientific data analysis but who was unaware of the powerful capabilities that Excel provides that is likely waiting just one Startup menu click away.
The book is not without fault. I would have preferred that it had been in color, or at least have one color section to show some of the more impressive visualizations that I'm sure would look great in color. In addition the index is silly short for a book that clocks in at 700 pages. But those are only minor quibbles for what is all-in-all an amazing piece of work.
You can purchase Advanced Excel for Scientific Data Analysis from amazon.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.
The first chapter concentrates on the getting the most out of Excel as a tool. All the chapters that follow dig into specific data analysis techniques. Chapters two, three and four are on least squares. Chapter five and six cover the analysis in the time domain including fourier transforms. Chapter seven covers differential equations. Chapter eight returns to Excel by digging in deeper into macros. Which leads into chapter nine, where we dig deeper into basic mathematical operations. Chapter ten covers matrix operations. And chapter eleven wraps it all up by giving you some spreadsheet best practices.
In University style there are also some exercises that you can do along the way if you want to tweak your brain pan a little more. To amuse myself I tried a few and I believe the book would have assessed my attempts 'wanting' if it had a voice to tell me.
Where most books like this would have several authors this book has just one; Roberte de Levie. This means that the tone, style and quality of the book is consistent throughout. A fact that you will come to appreciate as the book wades in ever increasingly deep data analysis concepts as the chapters roll on.
Though I would have preferred the book to have code samples in C#, I understand that the language of Excel is VBA and I guess I have to live with that. Thankfully VBA has come a long way and if you so inclined it would likely be easy to translate the code into C#, Java, or whatever else you like.
The fact that one person wrote the book left me wondering, "Who is this guy?" In my minds eye I kinda of figured he would look like one of those pulsing brain guys from Star Trek. Turns out he is a professor at Bowdoin College. And his fields of study include ionic equilibria, electrochemical kinetics, electrochemical oscillators, stochastic processes, and a whole lot more stuff that almost seems made up to sound impressive.
When this book isn't serving as an amazing reference for both Excel, scientific problem solving, or just insane equations it serves other purposes as well. It's a handy portable IQ test, as the count of pages you can grind through in one sitting, plus 90, is roughly your intelligence quotient. And if you fail at that you can always put a copy of the book, along with the Orange Bible, under your pillow and try to osmose your way to becoming the Kwisatz Haderach.
In all seriousness, this is a great book. It represents the kind of in-depth work and research we used to see in books that came out twenty years ago. Robert is to be applauded for his work. This is an excellent resource for anyone looking to do scientific data analysis but who was unaware of the powerful capabilities that Excel provides that is likely waiting just one Startup menu click away.
The book is not without fault. I would have preferred that it had been in color, or at least have one color section to show some of the more impressive visualizations that I'm sure would look great in color. In addition the index is silly short for a book that clocks in at 700 pages. But those are only minor quibbles for what is all-in-all an amazing piece of work.
You can purchase Advanced Excel for Scientific Data Analysis from amazon.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.
Don't do it!
If the recent financial meltdown has left you wondering, 'When does exponential decay function stop?' then I have the book for you. Advanced Excel for Scientific Data Analysis
So THAT's why we had a financial meltdown. All of those investment banks were doing their books and analysis with Excel 2007.
Just disrupt the deflector shield with a tachyon burst.
There's something hard to reconcile about the reviewer's obvious awe and the fact that the book was written by someone who thinks doing meaningful scientific data analysis in Excel is a good idea.
Talk about the wrong tool for the job. If you need to do any sort of serious data analysis, use R, not Excel.
Give me Classic Slashdot or give me death!
"The first chapter alone will have you in awe as you see the author turn the lowly Excel into something that rivals Mathematica using VBA, brains, and a heaping helping of fortitude."
Then why not just use Mathematica?
But it seems you have not seen Word 2007 equation editor. It's close, in both capabilities, and output quality, to LaTeX. Too bad the rest of Word sucks ass.
>> that would have been a nightmare to build with any other tools
Someone should tell this guy about SAGE http://www.sagemath.org/
You see, there is a fundamental problem in science and the problem can be summarized as this: how do you get the right results in order to optimize the grants that you receive. Spreadsheets are ideal for this purpose for two reasons. First of all, they are designed to handle financial data. This is great because financial data are what grants are all about. For example: will result X allow for a conference in Hawaii or California this year.
The other big reason to use spreadsheets is that they make data more maluable. Normal scientific tools make it difficult to micromanage the data that you acquire, partially because the people who produce that software have this mistaken notion that data has to be managed in a consistent way. So you're usually stuck doing the same thing to an entire dataset, and it's even difficult to treat different datasets in different way. But spreadsheets expose all of that data, so it is easy to tweak an observation here and a variable there to get the desired result to maximize your grant.
So you see, spreadsheets are a tremendously valueable tool for scientists. It is the best tool for the job.
So? What's so special about that? You can turn C, Fortran, or even assembly language into something that rivals Mathematica using brains and a heaping helping of fortitude. This is arguably a better deal, since you don't need the VBA.
--MarkusQ
Mistaken Identifiers: Gene name errors can be introduced inadvertently when using Excel in bioinformatics
------ merkur (4T] berlin . c0m
...everything looks like a snowglobe!
Hardcore data analysis in Excel is almost always a bad idea. You can almost always find a way to do it in excel, and you can almost always find a way to do it better, faster, and cheaper somewhere else.
R, MatLab, Mathemateica, Python/Numpy, SigmaPlot, and any number of old, well written, debugged and vetted numerical libraries written in C or Fortran. I've used all of these at various times to solve something that a co-worker couldn't figure out how to do in Excel.
I fit quick linear regressions in Excel. For *anything* else, there is a better choice.
-V-
Who can decide a priori? Nobody.
-Sartre
As a graduate student in physics, I have never seen a serious researcher use excel for data analysis. Nor for that matter, is it common to see a scientist using windows for the OS--all linux and mac OS. This is akin to writing a book about publishing scientific papers with office. Instead, learn LaTex... The only group of people who use excel for large data analysis are financial types and MBAs. Need I remind you how that turned out?
for scientific data analysis.
I know it is popular and many science and engineering faculty lazily encourage their graduate students to use it. However, something like matlab beats the crap out of excel any day. Spreadsheets tend to obfuscate relationships between data, require a lot more clicking (read human intervention) and waste time that could be spent thinking about the data, and are singularly unsuited for analysis of similar sets of data (a situation any scientist faces when he has to do a series of experiments).
Matlab might take sometime to initially write the scripts, but it is so powerful and extensible that no one in their right mind would want to use excel. If you are a slave to spreadsheets, get yourself a copy of Microcal Origin or Labplot.
Excel is especially unsuited to the task of preparing figures for scientific publications. The default formatting is at once wrong for the task and hard to change. Once you set your preferences in matlab (easy to do), you are set for life.
In my experience, excel is also rarely used for anything serious outside of US. Maybe its an indictment of how lazy, slow witted and easily misled our pool of talent is becoming.
When I worked in the semiconductor industry in the late 90's, Excel nearly cost us several hundred grand. It had "helpfully" autocorrected a code in the documentation for a mask used in one of our clock buffer chip products. Had the engineers not caught this mistake in the printout, the fab of the chip would have been botched. The engineers were mad as I recall because they would change the code and Excel would change it back. If you can't prove what your tool is doing, you don't get to use it is what they taught me in engineering school.
Is perfectly safe and trusty for that kind of work. Thats why we are using it here at the Large Hadron Co
SPSS has now become the standard data analysis package for quantitative studies in social sciences. It's very crappy software, and it wouldn't take a whole lot of augmentation to get Excel do what SPSS does.
The problem is that social scientists don't want to mess with the internals too much, and SPSS made for them a point and click interface - in effect, they out-Microsofted Microsoft. They charge an insulting $1500/copy and completely dominate the universities, so they're making good money.
They seriously need some competition.
What in the world are you talking about? :)
LaTeX is a markup language. You can express math with it, but it doesn't do anything for you in terms of analysis.
Excel is good for small data sets and quick looks at stuff - but painful to develop in.
Mathematica requires college-level calculus and linear algebra... not PhD stuff by any stretch.
Anyway, you left out Matlab - which is pretty awesome. Depending on what you are doing, there is also R, Maple, Minitab, MathCAD, yada, yada, yada. Lately I've been doing stuff in Python... SAGE is pretty nifty, and the NumPy/SciPy stuff is coming along well (it is included in SAGE).
W..w..W - Willy Waterloo washes Warren Wiggins who is washing Waldo Woo.
When I was a freshman in engineering school, my intro to engineering class required us to purchase a book similar to this. We were given two class periods to work with Excel, supervised by a TA. (it was considered a lab) I remember the assignment involved proving that sin^2+cos^2=1.
If you couldn't figure out Excel within those two class periods, it was recommended that you switched your major to business administration. The business administration school had a semester long class devoted to learning Excel.
One of our competitors trademarked the term "hypothesis". From now on, we will call them "boneheaded ideas".
You cannot be serious ..
"Excel 2007, like its predecessors, fails a standard set of intermediate-level accuracy tests in three areas: statistical distributions, random number generation, and estimation"
davecb5620@gmail.com
Python for scientific analysis,
Python is the solution I recommend for everyone who looks for tips on advanced Excel uses. Excel is OK if you just want some quick and dirty solution for a small problem, but if you have to go to the trouble of reading a book, Excel is clearly not the best solution.
For scientists and engineers who need something more than what Excel (and possibly Matlab) offers, I recommend starting with either A Byte of Python or Dive Into Python.
This is one reason the VB scripting turns out to be highly useful.
I recently had to do a project in VBA/Excel after years away from it, and it made me want to dig my eyes out with a spoon.
Don't ever write custom functions... ever. You'll thank me when you don't have to worry about whether or not they silently fail.
And once I had my whole spreadsheet corrupt for some reason... had to open on a Mac and re-save it. Then it worked fine on the PC again! Aye.
W..w..W - Willy Waterloo washes Warren Wiggins who is washing Waldo Woo.
don't forget Octave! (it's more-or-less a FOSS Matlab clone, and follows more closely to Matlab syntax than SciLab)
Another book in the same line is E. Joseph Billo's "Excel for Scientists and Engineers," Wiley-Interscience, 2007 ISBN 978-0-471-38734-3, including CD.
You may or may not agree with using Excel, but if you do, this book will help with roots of real and imaginary equations, ordinary and partial differential equations, matrices, and statistics.
Sometimes you just don't have the luxury of using Matlab, Spotfire, etc.
Live in the Future; It's Just Starting Now!
openorifice... if only we could come up with a new graphic for their logo...
I thumbed through the book but not impressed.
The author probably has used excel as best as any one can in doing the task he intends. But for most of people, the effort to acquire the skill by reading the book is not well-spent, since one can probably learn other tools which really intended for scientific analysis.
For statistic packages, R probably is much better, though I would prefer SAS. Try a huge data set (200MB, and put it in excel, your system will crawl before excel crashes, but in SAS, it will be really fast, and provide much more statistics. How much faith do you have in Excel's statistical function anyway?
I don't.
As for differential Equations, I would try matlab, if I really doing NA, I will choose netlib's packge anytime over Excel.
One can know Excel really well, and bend it to do all kinds of job, but in the end, it can only do that much. you probably can use bash to write a trading system, but why bother?
So according to the book, here's the recipe:
1. Write your data analysis software in VBA
2. Use the Excel cells, buttons, bells, and whistles for the I/O
3. Profit!
The math is actually irrelevant. Any computational mathematics book that respects itself uses pseudocode for the examples. If it is possible to program in one programming language it should be possible in any other language too.
I tried it too, although I wasn't nearly as crazy to do any numerical computations in VBA. I wrote the program in pure good old Fortran 95, wrote some VBA scripts to read the Fortran ASCII output, and set everything else up in Excel that my boss liked (I'm a chemical engineer). There you go: it's fast to program, fast to run, easy to maintain.
I would like to see anyone try to keep up with the Microsoft paraphernalia between VBA-Excel versions, if the whole thing is written in VBA. Not to mention the problems that I had with the locale when I tried to run the VBA code in a computer running a german version of Excel that had decided that the decimal point is there as a thousands separator and the comma was used as a decimal point. The setting for it in Excel was nowhere to be seen (I still haven't really figured it out. The central Windows setting seemed to have no influence on it, although I suppose it should) and 1.234 was then 1234 and 1.2E-02 was a character string. Oh, the pain... Thankfully, my *basic* Fortran part absolutely did not care, it just worked, and only the I/O needed to be reviewed.
Try to send the program to a customer without knowing what kind of Excel version he is running. We had to go as far as Office 97 just to be sure, and there was still the problem with the locale. After a year, the I/O was useless, but who cares? It was only 1% of the code.
I would still use Excel, but for nothing other than the most trivial tasks. There are wonderful libraries out there that work with Fortran and produce very nice graphs on the fly.
In my experience, excel is also rarely used for anything serious outside of US. Maybe its an indictment of how lazy, slow witted and easily misled our pool of talent is becoming.
I recently spent some time in Japan in a design group for a large Japanese company. I was showed the massive spreadsheet used to calculate power plant capacity and consumption. I almost cried. The whole sheet was based upon one large circular reference. Nobody understood it and it referenced steam tables through a plugin but didn't show the output of these calculations.
The US is not alone in using Excel for things it wasn't meant to do. Now please excuse me while I run a couple monte carlo simulations in Excel.
Even those who arrange and design shrubberies are under considerable economic stress at this period in history.
Feeding the trolls is one thing - you're holding out a ham and calling "Here boy!".
This comment is for entertainment purposes only. Any similarity to real insight or information is purely coincidental.
Look at all those posts saying "Excel is not the right tool for this" or "When all you have is a hammer...". The point was not grokked by those folks.
I'll lay it out for you, plain and simple:
This book is like installing a linux kernel onto a wristwatch.
We should be marvelling at the feat, not lambasting a tool that was "hacked" to do so much more than it is normally used for. If you can't appreciate that kind of work, maybe you should just stick to appreciating fine arts.
/.'s 10 Millionth
> It had been years since I had seen a book typeset using LaTeX.
The publishing industry (including my company) typesets books using LaTeX all the time. The reason you don't notice it (apart from the superior quality) is that it does its job of typesetting very well.
If this book has been typeset using LaTeX then I'm a Dutchman, or something has gone very wrong (and I'd like the author to contact me to let me know what).
Perhaps he was given faulty fonts, perhaps he was using a badly-written publisher's style, or perhaps he -- or his editor -- spent a long time making it look as bad as possible. Maybe OUP had it completely re-typeset in some other system without telling him. There are at least a dozen typographic faults in one paragraph alone, from unnecessary hyphenation to excessive word-spacing to bad math spacing, and LaTeX simply doesn't make those types of mistake unless you work very hard to introduce them manually.
As a test I screenshot a random paragraph that I viewed in Amazon's "Look Inside" feature, and then retyped it in LaTeX and typeset it (PDF).
As I don't have the book (and wouldn't understand it anyway :-) I'd be interested to know where the information came from that it was typeset with LaTeX; and if it really was done in LaTeX, I'd love to know WTF kind of style files, fonts, and preamble were used.
Maybe some of the people yelling about how Excel is the wrong tool can give some advice for my scientific data analysis and visualization needs.
I have simulations (written in C++ and Python) that spit out tab delimited data files. I then need to analyze that data, doing things like linear regression on subsets of the data and calculations to transform the raw data into something else for plotting.
I have a Mac (with Windows XP in Parallels), I am not a student, and I don't have much budget ($500) for software. Currently I use a Mac program called Plot which is a little buggy and incomplete but has some nice plotting abilities. When I need a spreadsheet I use Apple's Numbers, but that seems sorely limited in abilities. What's a better tool for this job?