Traditional graphics package takes a right turn, bulking up on analytic capabilities while becoming even easier to use
OriginLab just took a right turn with their flagship product and bulked up the analytic capabilities while retaining the graphics and making the product easier to use.
click to enlarge
Figure 1: Gaussian data workbook.
From previous reviews of Origin, the reader knows that this software was useful for graphics and limited statistics. Version 8 has an enhanced statistical repertoire, and OriginPro 8 has all of these features plus more analytic tools, 3-D fitting, image processing, spectroscopy (baseline and peak fits), and signal processing. It also possesses the ability to automatically recalculate analytic results. Actions may be issued from the menu-driven elements or from a command line. Table 1 lists many of these new features for statistics; and Table 2, for image processing.
For the novice, help is available from a variety of sources: the manual, the software, the Web site and the technical support group. The manual is supplied with the software and is usually clear, concise and well organized. As usual, the index is all too brief, which limits its use and forces the user to the Help section on the toolbar. Not to worry, however, as this feature is very complete and full-featured. I found the Search function to do a good job on the first try. Although the image processing features are not covered in the manual, they are in the Help section. A brief run through the tutorial chapter is highly recommended for new users.
Windows, workbooks, worksheets and graphs
Workbooks are used to organize the data and each is composed of one or more worksheets which, as with most modern programs, are Excel-like to flatten the learning curve and provide useful tools. To demonstrate the ease-of-use, we can run through a simple example from the manual.
The experimenter has some data from a near-normally distributed (Gaussian) phenomenon and wishes to graph the data with appropriate error bars. To begin, we create a new workbook with the
click to enlarge
Figure 2: ScatterPlot of the Gaussian data
Toolbar selections File:New. We then import the data with File:Import:Single ASCII to bring up the open dialog box. We choose the Samples/Curve Fitting folder and highlight the Gaussian file. Clicking the 'open' button then brings up the data (Figure 1). The log dialog box that also appears may be hidden, reduced or docked.
Notice the feature called Sparklines. This appears in one of the title cells and allows the user to immediately view the shape of the data. Also, the sheet name automatically becomes the file name. Now for the graph…
First, highlight the three columns in the worksheet, and then select Plot:Symbol:Scatter from the toolbar menu to create Figure 2. In this case, note that the data and error have been appropriately plotted. We informed OriginPro that the last column was to be used for error by simply clicking that column's title cell and then right-clicking to bring up a fly-out menu to select Set As:Y Error. As with many modern programs, we have a navigator pane (called a Project Explorer) to quickly move between worksheets and graphics. The developers sometimes see fit to introduce new nomenclature such as the 'graph layer.' This is defined as the axis-scale values and data plots, as well as any included text labels, drawing objects, legends, colors, button objects, etcetera. Fortunately, these are kept to a minimum and well-defined.
In the above example, we merely imported pre-cooked data, so the question arises as to how simple it is to import one's own data. By paying careful attention to the format and placement of column labels, this is actually easy to do with just a little practice. The software provides both an
click to enlarge
Figure 3: Simple column statistics
Import Wizard and pre-defined formats (e.g., Excel, MATLAB, Minitab, SigmaPlot, KaleidaGraph and more) to simplify the task.
Once accomplished, it was straightforward to generate simple column or row statistics to verify data, identify problems, and further snoop (Figure 3). These are added as a separate tab to the data worksheet. The choices for these statistics are ample and will give the experimenter a nice snapshot of the data distribution.
Graphics and layout pages may be saved and exported in 17 common image file formats.
OriginPro 8 does a variety of statistical tests running the gamut from simple row and column descriptive statistics to hypothesis testing, ANOVA, survival analysis and power/sample size studies. This last is taking on ever more importance, as both journal editors and government regulatory agencies are increasingly demanding some justification for the experimental design.
click to enlarge
Figure 4: 1-way ANOVA
The statistical tests are quite a bit easier to run than the graphics (a lot less formatting) but, in some cases, there is still work to do. For example, Origin likes stacked data for the ANOVA, i.e., one column for class designators and one for data. As I could find no easy way to stack independent column data, this was done in Excel prior to importation. Once accomplished, OriginPro 8 will produce some very nice diagnostics and plots in the results window (Figure 4).
Some of the functions with which I was less familiar were also tested and found to be simple-to-use with the help of the dialog boxes. These boxes will suggest the data format so that the user can edit the data prior to the analysis and get a smooth execution the first time.
Although OriginPro 8 has many new and extended features, and the software gets easier to use with each version. I still would like to see a more seamless integration with Windows Vista (all of the problems that I encountered were Vista-driven rather than OriginPro problems). There are several 3-D types that still require specific matrix transforms and some actions that are still not quite as intuitive as they should be. However, as with all software, time spent in practice is well-rewarded. For the price, OriginPro 8 has many features that recommend it for many applications not restricted to the natural sciences.
The Web site contains concise information that is easy to access. Readers are encouraged to download and test an evaluation copy.
• $1,500 commercial
• $700 academic
One Roundhouse Plaza, Suite 303
Northampton, MA 01060
John Wass is a statistician based in Chicago, IL. He may be reached at editor@ScientificComputing.com.