“JMP is my go-to tool for on-the-fly data analysis.”
- Kaiser Fung
- Statistician and author of Numbersense
Import and Process Data
The most time-consuming part of any analysis is importing and reshaping data, which can come from a variety of sources. JMP excels at preprocessing, even with large data sets, and can import data from most popular file formats and locations, including SAS and ODBC-compliant databases.
If your data lives in Excel spreadsheets, the Excel Import Wizard lets you easily extract raw data from multiple sheets in various formats to prepare it quickly for analysis. JMP also makes short work of sampling from huge text files. Control of this sampling allows you to explore the data quickly with files that otherwise would be too large to hold in memory on your desktop computer.
Data manipulation and organization are almost always a necessary part of any analysis. You can easily join, concatenate, transpose, split, compare, stack, sort tables, or create a table subset, all without programming. A drag-and-drop Tabulate tool lets you quickly group and summarize a table, producing a preview of the output that updates automatically.
Once your data is in a JMP data table, you can quickly pose and address specific analytic questions, dynamically recoding, regrouping and making new variables if needed.
In addition to the drag-and-drop, point-and-click tools, JMP gives you the option to automate boring and repetitive tasks without coding, saving you time and letting you get more done faster.
Spreadsheets don't easily reveal patterns and trends in data sets. And seeing patterns helps you make discoveries. JMP provides rich and dynamic visualization tools, making statistical discovery easier and more effective, leading to innovation. In nearly every analysis platform, JMP provides comprehensive graphical output that lets you convey important findings to decision makers with clear, concise and compelling visualizations.
The patented Graph Builder is often the best way to begin exploring your data. Interactively build simple or complex graphical displays, including polygon region maps, just by dragging and dropping. Simply drag the required variables into position, choose the required graph element from an icon palette, and customize the display to get the final result.
You can add background maps to all relevant JMP graphs using high-quality, built-in geographic images, or plot data on street-level maps that include features such as cities, roads or bodies of water. With Bubble Plot, you can create animated data movies, showing changes in many dimensions over time.
Dynamic linking capabilities perhaps best illustrate the magic of JMP. You can select rows from any graph and instantly see the selection propagate to multiple views, yielding insights that other software simply can’t reveal.
When you are ready to communicate findings to others, your options are equally diverse. Share JMP data tables to Graph Builder for the iPad® and explore graphs on the go. Or generate interactive output of many JMP reports that can be viewed quickly in any device with a browser. And, of course, JMP can create presentation- or publication-ready graphics in a multitude of file formats.
Basic Data Analysis
Employing basic tools for visual analysis is often the best way to communicate results and motivate action in an organization. And frequently, your first step in a statistical data inquiry consists of investigating variables one at a time, known as univariate analysis. In JMP, once you've identified the columns that interest you, Distribution automatically provides graphs and statistics based on the variable's defined modeling type.
Quickly get histograms, summary statistics, box plots and quantiles for continuous data, capability analysis, distribution fitting and frequencies for nominal or ordinal values.
- Key capabilities in JMP for univariate statistical analysis include:
- Descriptive statistics.
- Statistical tests.
- Box plots.
- Distribution fitting.
- Statistical calculators and simulators.
- Basic inferential testing.
- Correlation and regression.
Group and Filter Data
In any business, the quicker you can learn and adapt to ever-changing customer needs, the quicker you can get ahead of your competition. To accelerate this learning cycle, you need to be able to notice patterns in your data, focus on the most important, and act quickly. You can’t waste time generating a stack of reports to wade through, or worse, writing custom code and waiting for output before acting.
JMP brings a radically different approach to the daily task of slicing and dicing data. Its grouping and filtering paradigm allows for instant in-memory recasting of report output. Imagine how quickly you can focus on interesting areas when you can update reports on the fly just by clicking through levels of a categorical variable. With one click, you can even switch the analysis focus to a new metric entirely.
- Grouping and filtering tools in JMP include:
- Local and global data filters for focusing on specific parts of your data table, with or without conditional statements. The ability to save favorite filter settings brings efficiency to routine filtering tasks.
- Easy-to-define row markers, colors and labels that enrich graphical reports and data tables.
- Column Switcher for swapping variables within a graphical or statistical report. Stepping through variables manually or by animation allows you to spot patterns and anomalies when you have hundreds of variables.
- Transforms for generating derived variables on the fly. Stay in the flow while you are analyzing data and create many statistical or mathematical transformed columns of your data with a single click.
Design of Experiments
Many organizations rely on “A-B testing” for experimental design, but testing one situation against another with many factors in flux is a very slow way to learn about your business.
In contrast, design of experiments (DOE) in JMP offers a proven and practical approach for exploring and exploiting the multifactor opportunities that exist in almost all real-world situations. Using multifactor experiments, you learn more quickly, at minimal cost, by teasing out not just the effect of an individual factor, but also the combined impact of two or more factors. JMP offers leading-edge capabilities for design of experiments, so you can design the best experiments to answer specific questions. JMP also offers a rich set of analyses tailored to your design in a form you can easily use.
With JMP, you don’t have to fit your problem to a textbook design and risk wasting resources. The unique Custom Designer constructs a design to fit your problem, taking into account specific conditions such as time, budget and other experimental constraints.
Many analysis problems include hard-to-change variables, such as the temperature of a reaction vessel or the location of a cornfield. A completely randomized design would require such factors to be reset after each experimental run, which is clearly impractical or cost-prohibitive. The design most appropriate in such situations is a split plot, and JMP can generate I-optimal split-plot, split-split-plot and strip-plot designs. JMP also includes the correct random-effect restricted maximum likelihood (REML) model in the table that contains the experimental worksheet to make the analysis rigorous but also straightforward. No other commercial software package offers this level of flexibility with split-plot designs.
In addition to Custom Designer, JMP also supports classical screening (fractional factorial), response surface, full factorial, nonlinear and mixture designs, as well as advanced designs, including accelerated life tests and designs for computer simulation, such as cluster-based, space-filling designs that allow for inequality constraints on factors.
Also, JMP is the only software that implements definitive screening designs. The most important new class of designs in the past 20 years, definitive screening designs are used to efficiently and reliably separate the vital few factors that have a substantial effect from the trivial many that have negligible impact. A traditional screening design may erroneously screen out a factor that actually has a strong curved effect. And if there are two-factor interactions, traditional screening designs will require follow-up experimentation to resolve the ambiguity. Neither of these limitations exist when using definitive screening designs.
By making a useful separation of data into signal and noise, statistical models encapsulate trends and patterns so you can better learn about your business, your competition and your customers. With this knowledge, you are empowered to take the best course of action and grow your business more easily.
Building useful models is part science and part art, and JMP includes an array of statistical platforms to help you build useful models of your data. With methods for revealing relationships among variables in a process, JMP allows you to not only make predictions but also to identify settings for factors that yield the best performance. JMP includes a variety of linear and nonlinear ways to fit models, and these diverse fitting tools help you make correct decisions, whatever relationships your data shows.
At the heart of the JMP model-fitting toolkit is the Fit Model platform. Fit Model lets you construct model terms and select from different methods, including standard least squares fitting, stepwise, MANOVA, generalized linear, loglinear variance or logistic regression (nominal and ordinal). JMP also fits models with REML and includes advanced multivariate modeling techniques: principal components, partial least squares, cluster, item analysis, partition models and more.
The Fit Y by X platform is intended for modeling dependencies between a single input and a single response or outcome. This platform supports simple linear regression, logistic regression, ANOVA, ANOM and contingency analyses. It responds to the modeling types of the variables used, and so unifies many commonly used methods if you don't need the control that Fit Model offers.
Too often, extracting meaningful conclusions from messy data entails the arduous process of exploring one variable or response at a time to find the key drivers. And when your data has missing values or outliers, you can easily find yourself in trouble. JMP helps you build models with many columns by providing automated options for response screening to investigate the relationships between many X and Y variables. Also, modern significance adjustment techniques – such as the false discovery rate – help you draw useful conclusions.
The JMP Nonlinear platform makes it easier to fit nonlinear models to your data. The software's built-in library makes it particularly simple to fit popular bioassay and pharmacokinetic models. By defining appropriate formula columns, you can fit virtually any nonlinear model.
Advanced model-fitting capabilities in JMP include neural network models (the Neural platform) and decision trees (the Partition platform).
No matter what modeling techniques your data requires, you can quickly and easily build useful models in JMP.
Building models is useless unless you can motivate change. JMP lets you communicate model results and perform what-if scenarios visually to understand the pattern of predicted response and how each factor affects it. In JMP, you explore this opportunity space visually through the Profiler, a dynamic tool for interacting with your model no matter how it was built. The Profiler lets you grab a factor, dynamically adjust its level and immediately see the impact on the response in a way that a table of calculated coefficients would never allow.
- With JMP, you can also:
- Set up desirability functions for responses and then find the optimal settings of factors that maximize outcomes across the responses.
- Use Monte Carlo simulation techniques to assess risk when inputs are uncertain.
- Fill an opportunity space that you can then filter to show the feasible region.
- Use other, more specialized profilers to visualize the predicted response.
Preventing failure and improving warranty performance are two of the most important reasons for using proven techniques to fully understand the performance of your products over time. JMP helps you pinpoint defects in materials or processes; it also helps identify design vulnerabilities so you can understand how best to correct them.
Do you need to determine the most appropriate distribution to use for making reliability lifetime predictions on your products and components? Let JMP automatically evaluate a large range of reliability distributions to find the best fit. Using Life Distribution analysis in JMP, you can specify a nonparametric distribution as well as 19 parametric distributions and compare fits visually.
- JMP includes a rich set of capabilities for reliability analysis:
- Fitting life distributions.
- Fitting life distributions with one factor (e.g., accelerated failure models).
- Performing recurrence analysis.
- Modeling product degradation.
- Estimating survival, parametric survival, and proportional hazards models.
- Designing accelerated life test (ALT) experiments.
- Performing Crow-AMSAA analysis of repairable systems.
- Forecast warranty returns from failure data.
Quality and Process Engineering
The market demands continual improvement, which is why you strive to accelerate time to market, protect your brand by minimizing customer complaints, and deliver products and services that consistently meet or exceed customer expectations. JMP has the necessary tools to be at the heart of your quality program, providing a wide range of relevant graphical and statistical capabilities.
You can monitor processes with the full set of control chart types included in JMP, or build control charts interactively with drag-and-drop tools in the unique Control Chart Builder. With a workflow analogous to Graph Builder, Control Chart Builder lets you perform what-if analyses with your process data and explore many subgroup and phase variables and their effects on your processes. You see problems in ways that are impossible using static control charts.
The Measurement Systems Analysis platform supports multiple analysis methods, including Donald J. Wheeler's evaluating the measurement process (EMP). Other quality analysis features in the software include provisions to perform Gauge R&R studies and create Pareto charts. You can easily visualize sources of variation in your measurement process, evaluate product defects, and monitor processes for stability. JMP also lets you investigate out-of-control conditions and perform root-cause analysis.
Consumer and Market Research
No matter what role you play in it, marketing is complex and rapidly evolving, driven by the influx of digital technologies. Yet key business issues endure: the need to find the most profitable growth opportunities, develop the best products and services, take the best marketing action, and maximize cross-business impact. To stay competitive, you need to socialize your brand, listen to customer feedback and then adapt your products and services. Whether you are conducting exploratory, descriptive or causal research using primary or secondary sources, JMP provides a comprehensive repertoire of tools for quickly and easily getting value from metric and nonmetric quantitative data.
- Capabilities for performing consumer research in JMP include:
- Categorical (survey) data analysis.
- The ability to import data in many external file formats, including SPSS Desktop Reporter.
- Single-click analysis of simple, related, multiple and structured response survey questions.
- Choice experiments to optimize design of your goods based on consumer feedback.
- Factor and item analyses.
- Segmentation and clustering (k-means and hierarchical).
No matter what your level of statistical expertise, JMP will help you find new consumer insights more quickly and allow you to communicate findings to other stakeholders to drive consensus and action.
Extensibility via Scripting
Buying software that cannot grow past your initial needs leads to early obsolescence and expensive replacements. JMP includes many basic and advanced ways to mass-customize, even extend the software to address the unique challenges that arise as usage grows and your organization evolves.
The rich JMP Scripting Language (JSL) lets you work interactively and then save results for reuse. Power users can develop new functionality to solve problems that core JMP does not address. These custom scripts can even integrate capabilities from other applications.
Use Instant-App tools in Application Builder to create custom displays from multiple reports and graphs in a drag-and-drop interface. Application Builder’s design-time development environment also lets you create complex analytic applications by simply dragging and dropping. And when you finish your custom application, deploy it to other JMP users in your organization with the easy-to-use Add-In Builder.
After writing scripts, use the full-featured JSL debugger to interactively step through each line of code, examining variables to determine what might be preventing a script from performing its intended function. And use the JSL Profiler to fine-tune a script’s performance by monitoring the time it takes to complete each step of a routine and iteratively optimize your code for peak efficiency and speed.
It's easier to work productively if you can configure your software to work the way you think. Consistent settings, graph output and even color palettes mean fewer steps to understanding data. JMP gives you a comprehensive set of preferences that enable you to control fonts, graphic options and detailed settings within platforms.
You also can choose to display only those analytic tools and menus you use routinely.
JMP® as an Analytic Hub
As a member of the SAS family, JMP offers a seamless interface to the unparalleled capabilities of SAS. The deep analytics, reporting and data management capabilities of SAS extend capabilities of JMP desktop software to the server and beyond. You can also use JMP with other analytic tools, including a full interface to the power of MATLAB, complete calling interface to Dlls and the rich set of specialized libraries in R. JMP makes it easy to reach out to these resources and bring back results for dynamic data visualization and analysis.
Does your data live in spreadsheets? With the JMP Add-In for Microsoft Excel, you can easily move data from Excel into JMP or bring the power of the JMP Profiler to your spreadsheet models, enhancing the data with the advantages of JMP visualization.
Key Features of JMP®
- Ease of use
- Spreadsheet view of data.
- Desktop application, in-memory data analysis.
- User-friendly, point-and-click interface.
- Dynamically linked graphics, data and statistics.
- Integrated manuals and context-specific help.
- Modules for learning statistical techniques.
- Interactive hover help.
- Drag-and-drop formula editor.
- Markers, row states, row colors.
- Extensive customization preferences.
- System capabilities
- JMP files, scripts shared between Mac and Windows installations.
- SAS integration.
- Integration with R and MATLAB.
- Presentation features
- Journaling for saving analysis or presentation.
- Layout menu item.
- Drawing tools.
- Instant-app drag-and-drop reporting with Application Builder.
- Copy/paste reports to Microsoft Word or PowerPoint.
- Export publication-quality vector images (EMF, EPS, SVG), 72- or 300-dpi bitmap graphics files (PNG, JPEG, TIFF), PDFs and more.
- HTML export.
- Interactive HTML with data report output.
- Send JMP data tables with Graph Builder scripts to Graph Builder for the iPad® for exploration.
- Basic statistics
- Descriptive statistics.
- Confidence intervals, chi-square, one- and two-sample t-tests, paired t-tests.
- Correlation and covariance.
- Match paired data.
- Normality test.
- Equal variance test.
- Sample size and power for testing proportions or means.
- Regression analysis
- Linear regression, linear regression in Graph Builder.
- Nonlinear regression.
- Polynomial regression.
- Spline fits.
- Logistic regression (nominal and ordinal).
- Multiple regression.
- Residual plots.
- Statistical process control
- Shewhart charts (continuous and attribute).
- Control charts: XBar, R, S, XBar-R, XBar-S, EWMA, CUSUM, c, u, IR, p, NP, UWMA, and more.
- Rare event charts (g- and t-charts)
- Pareto charts.
- Capability analysis.
- Drag-and-drop control chart building.
- Tests for special causes.
- Measurement systems analysis (MSA). Supports Wheeler's EMP, Gauge R&R, and variance components methods.
- Variability charts.
- Cause-and-effect diagrams.
- Reliability analysis
- Fitting of life distributions, time-to-event.
- Accelerated life testing (Fit Life by X).
- Recurrence analysis of repairable system.
- Reliability forecast and growth.
- Kaplan-Meier survival plots and life tables.
- Proportional hazard model fitting.
- Plots: distribution, probability, hazard and survival.
- Multivariate analysis
- Principal component analysis.
- Discriminant analysis.
- Cluster analysis and dendrograms.
- Factor analysis.
- Correlations and scatterplot matrix.
- Partial least squares (PLS).
- Time series and forecasting
- Time series plots.
- Moving averages.
- Auto-, partial auto- and cross-correlations.
- ARIMA and seasonal ARIMA.
- Fit ARIMA group.
- Smoothing models.
- Winter's methods.
- Spectral analysis.
- Generate SAS code.
- Nonparametric tests
- Van der Waerden.
- Multiple comparisons.
- Spearman’s, Kendall’s, Hoeffding’s and more.
- Reporting tables
- Correspondence analysis.
- Drag-and-drop summary tables with Tabulate.
- Contingency tables.
- Fully featured JMP Scripting Language (JSL).
- Record, repeat, program, automate and customize tasks.
- Save analysis to script window or data table.
- Built-in SAS Editor: write or create SAS code in JMP, submit to SAS and view the results in JMP.
- Debugger and profiler to optimize scripts.
- Application Builder for custom applications and GUIs.
- Add-in Builder for easily deploying scripts as JMP add-ins.
- Data and file management
- Import/export: Excel, delimited and fixed-width text files, HTML, SAS.
- Import with best guess/preview.
- Excel import wizard.
- Open R code, MATLAB code, SPSS data files, Minitab worksheet files.
- ODBC querying.
- Open Microsoft Access databases, Teradata databases, xBase and dBASE files.
- Excel add-in option.
- Matrix algebra.
- Data manipulation: summary, subset, sort, stack, split, transpose, concatenate, join update, change data type.
- Data filter/local data filter.
- Drag-and-drop crosstabs/pivot tables (Tabulate).
- Missing data pattern.
- Compare data tables.
- Technical support and documentation
- Award-winning, free 24/7 technical support provided by SAS.
- Searchable online documentation on jmp.com and PDFs available in the software.
- More than 400 sample data sets with analysis scripts.
- Menu cards, quick reference cards and other resources available on the JMP user community website.
- Learning library, case study library, interactive learning tools and calculators.
- Graph Builder: Drag-and-drop graphing.
- Bar, line (run charts), needle, point and pie charts.
- Frequency distributions.
- Scatter plots, box plots, histograms, time series plots, contour plots, overlay plots, normal quantile plots.
- 3-D scatter plots.
- Special-purpose graphs such as ternary and variability charts, N-dimensional outlier detection, leverage and interaction plots.
- Normal plot, Pareto plot, cube plot.
- Interactive profilers (factor, contour, surface).
- Distribution fitting.
- ROC curves.
- Treemap, cell plot, parallel plot.
- Mosaic plot, stem-and-leaf plot.
- Mapping support: built-in shape files for US and world maps. Custom map shapes also supported.
- Street-level maps
- Bubble plots.
- Analysis of variance (ANOVA)
- ANOVA and ANCOVA.
- Unbalanced nested designs.
- MANOVA and MANCOVA.
- Fully-nested designs.
- Analysis of means.
- Multiple comparisons.
- Residual, main effects and interaction plots.
- Power calculation.
- Statistical modeling (fit model)
- Mixed models with REML.
- Repeated measures analysis.
- Stepwise regression.
- Generalized linear model (GLM).
- Response surface regression.
- Log Variance.
- Monte Carlo simulation.
- Scaled estimates.
- Durbin-Watson statistics.
- Box-Cox transformation.
- Cook's D Influence, hats, studentized residuals and more.
- Save prediction formula.
- Parameter power.
- Effects tests and custom tests.
- LS Means tests and contrasts.
- Generate SAS code.
- Design of experiments
- Design optimal experiments using the Custom Designer.
- Definitive screening design.
- Split-plot designs: Split-plot, split-split, strip-strip using REML.
- Response surface design.
- Full-factorial design.
- Fractional-factorial and screening designs.
- Choice design.
- Space-filling design.
- Accelerated life test design.
- Nonlinear design.
- Taguchi arrays.
- Augment/evaluate design.
- Sample size and power.
- Plots: factor profiling, residual, main effects, interaction, cube, contour, and more.
- Exploratory data mining/predictive analytics
- Stepwise, all possible models and model averaging.
- Recursive partitioning (classification and regression trees).
- Neural networks.
- Missing data handling.
- Built-in cross-validation (k-fold and holdout).
- Consumer and market research
- Survey analysis, distribution of responses.
- Choice/conjoint design and analysis.
- Factor analysis.
- Item analysis.
- Share/frequency charts.
JMP runs on Microsoft Windows and Mac OS. It includes support for both 32- and 64-bit systems.