Data Visualization in Excel: A Guide for Beginners, Intermediates, and Wonks

Who could have imagined that Excel could be this awesome? Reading Jonathan Schwabish's book about Data Visualization in Excel is akin to uncovering secret abilities within your tried and true spreadsheet program. He not only instructs on creating attractive charts but also teaches how to convey narratives using data.I adored the way he guides you through his experience, starting as a naive college student and becoming an expert in data visualization. It's similar to having a helpful advisor leading you through the labyrinth of spreadsheets. What is the top aspect? He always stays authentic. No complex terminology or confusing equations. Simple and useful advice in everyday language.If you've ever found yourself bewildered by a blank Excel spreadsheet, this book is here to help you. It's similar to discovering a hidden code that unleashes all the possibilities of your data. Believe me, once you read this, you'll see Excel in a completely different way.

MS Excel remains one of the most widely used programs for working with data, including visualization. Some people use Excel and little else, while others use it together with more advanced or more versatile software. The aim and achievement of Jonathan Schwabish's book is to show how you can go beyond the basic menu of graphics possibilities in Excel to produce a variety of other displays, most now fairly standard and others more unusual. The author's style is friendly and indeed conversational, all the way down to ahem, yikes, yeah, terrific and incredibly. (He does treat "criteria" as singular and will irritate linguistic fuddy-duddies by confusing "compared to" and "compared with".) More importantly, this book is packed with tiny technical tricks that would take much experimentation to discover for oneself.Possible readers vary from people whose primary aim is communication -- so that their main job is to visualize data directly, whether for colleagues at their workplace or more publicly -- to people whose primary aim is research -- so that visualization is part of some more elaborate analysis. Schwabish has written previously for both kinds of readers. This range is tied up with how far people are interested in statistical summary or indeed statistical modelling. My own prejudice is that researchers interested in visualization are better off using something more versatile, but that discussion goes way beyond this review. Similarly, some of the trickery here looks tedious and even error-prone, but that may be acceptable pain for anyone unable or unwilling to devote much time to learning something quite different.The assumptions here about readers run essentially that they have some Excel and visualization experience, so this book doesn't start at an absolute beginning of explaining say bar or line graphs or scatter plots. Despite some passing mentions of regression, percentiles, quartiles, and quintiles, readers are not taken to be familiar with even introductory statistics.On the graphical side, experienced readers won't all agree on which plots should be covered, but most of the choices here are modern standards. I missed some basic uses of scatter plots, including plotting different groups of points, convex hulls, or smooth summaries; moving averages or trend fitting for time series; quantile (rank-size) plots; and use of logarithmic scale, one of the best devices for seeing order within messy or awkward data. Other way round, I've never seen the point of bullet charts or how stripe charts improve on line charts. Frequency polygons, here introduced as "overlaid area charts", are all too likely to confuse, as they misrepresent binned frequencies by segmented lines. The raincloud plots in Ch.28 are a little hard to like, if only because the whiskers are placed unconventionally.Most of the advice is again standard. There is much helpful detail on use of colour. The simple point that orange and blue make a good pair, with grey as a complement if needed, is well shown in examples. A rule that bars should always start at zero (p.39), although often given in graphics books, is too emphatic. I agree that bars encode distance from a reference level. That reference is often zero, but may sometimes be another level, say freezing point as 32 degrees Fahrenheit, a mean value, or parity, a ratio of 1:1. Similarly, although a neutral category is sometimes omitted from diverging bar charts, any idea that you should do this and only ever plot its frequency separately (p.154) is a trifle dogmatic.Some small errors can be seen over units of measurement. Thus various time series for eggs, fish and meat consumption in the United States are explained as in pounds per year (pp.7-9), which should also be per capita. Per capita is right on p.175, but not the factor of a billion which has somehow crept into the text. Units of thousand US dollars (presumably) are missing on p.51, while the use of Fahrenheit needs more flagging at several points, as the United States is just about the only country in which it is a standard scale.I liked the direct labelling (text by graph elements, not presented in a legend or key) at pp.7-9, but this was also needed on p.328. Excellent hybrid graphs and tables can be seen on pp.94, 322, 336 and 338, and that idea deserved even more exposure. One of the merits of the design here called broken stacked bars is that annotation with numeric text could be added helpfully. Alaska is missing from the maps on pp.332-333. Note that ggplot, meaning here ggplot2, is indeed very popular among R users, but it is by no means the only way to produce graphics in that software.