Summary of what was done in the 30-minute omnibus seminar "JMP Information Bureau" (5th to 8th sessions)
Last Modified: Apr 25, 2024 2:08 AM
From March to April this year, the JMP Japan Division held a seminar called "JMP Information Bureau" every Tuesday for a total of eight times. The purpose of this seminar was to convey various information about JMP in a short period of 30 minutes each time, and many more people participated than expected.
Many reactions to the ending
This blog summarizes the main topics covered in the 5th to 8th sessions, which were held in April.
*Please refer to the following blog for the main contents of sessions 1 to 4.
table of contents
- Episode 5 (Original broadcast 4/2, rebroadcast 4/9) Learn again what can be done with "univariate distributions"
- Episode 6 (original broadcast 4/9, rebroadcast 4/11) "About prediction intervals and tolerance intervals"
- Episode 7 (original broadcast 4/16, rebroadcast 4/18) "AI Awareness Survey" Analysis
- Episode 8 (Original broadcast 4/23, rebroadcast 4/25) Useful functions (Tips) recommended for JMP beginners
Episode 5 (Original broadcast 4/2, rebroadcast 4/9) Learn again what can be done with "univariate distributions"
We introduced various options for "Univariate Distribution," a platform frequently used by JMP users.
●Options for categorical variables (nominal scale, ordinal scale)
When you open the JMP sample data "Employee Taste.jmp," select [Analyze] > [Univariate Distribution] and specify the nominal scale columns "TV" to "Restaurant" as [Y, Column].
By clicking the red triangle to the left of the report column names, you can specify options to make your report look like this:
This report can be created by applying the following options in sequence:
*Hold down the Ctrl key, click the red triangle button to the left of the column name, and select an option to apply that option to multiple reports of the same format.
Options for continuous variables
In the fourth installment, “Analysis of Taylor Swift’s Songs,” we looked at a continuous scale column showing the popularity of each song.
For histograms in continuous reports, you can use the Tools > Hand Tool to change the histogram bin width and tick placement.
If you suspect a multimodal distribution, click the red triangle to the left of the column name and select Fit Continuous Distribution > Full Fit , then examine the Distribution Comparison report.
In "Compare distributions," the distributions are sorted in order of smallest AICc (better fit). In this example, the binormal mixture distribution and the trinormal mixture distribution show better fit than the normal distribution.
In cases like this, it may be a good idea to consider whether there are stratification variables that separate the distributions.
Episode 6 (original broadcast 4/9, rebroadcast 4/11) "About prediction intervals and tolerance intervals"
I explained the general idea of prediction intervals and tolerance intervals used in statistics, and demonstrated them using JMP's "Univariate Distribution." The handouts and sample data for JMP used in this lesson are attached to this blog.
Source: JMP Information Bureau 6th Prediction Interval and Tolerance Interval.pdf
JMP sample data: JMP Information Bureau Part 6: Fictitious data for explaining prediction intervals and tolerance intervals .jmp
Episode 7 (original broadcast 4/16, rebroadcast 4/18) "AI Awareness Survey" Analysis
We conducted an "AI awareness survey" among participants of this seminar and demonstrated an example of analyzing the data using JMP.
Example of a question
112 people responded to the survey. In addition to questions about AI, the survey also included questions about gender and age (over 40, under 40).
In analyzing the survey, we used the “categorical” and “latent class analysis” platforms in particular.
● "Categorical" platform
Using [Analyze] > [Consumer Research] > [Categorical] , you can efficiently analyze categorical data (nominal scale, ordinal scale).
In this survey, we asked participants to rate six items (a to f) about their impressions of AI using the same five-point scale from "applies" to "does not apply." For questions like this, we set them as [Responses with Common Values] as shown below.
Furthermore, specify "gender" and "age" in [X, grouping category] to compare the responses when divided by gender and by age.
This is a categorical report. The share chart at the top shows the share chart for men and women for each item specified as a common response.
When you select "Test for homogeneity of responses" from the red triangle button in the upper left, the "Test for homogeneity of responses" will be displayed at the bottom of the report. In this example, the chi-square value and p-value for the null hypothesis that "there is no difference in the response rate between men and women" are displayed, and the item "Enriching life"Only is significant.
● "Latent Class Analysis" platform
Analyze > Cluster Analysis > Latent Class Analysis allows you to separate respondents into groups (clusters) using categorical responses.
Here, we use latent class analysis to divide respondents into two groups (clusters) using gender, age, enriching life (combined into three categories), and being familiar (combined into three categories).
The interpretation of clusters 1 and 2 is based on the difference in their membership distributions.
Cluster 1: Positive about AI
Cluster 2: A group consisting of mostly men over 40 years old who are somewhat negative about AI
Furthermore, you can save which respondents belonged to which cluster in a data table by clicking the red triangle button to the left of the report "Latent Class Analysis (number of clusters: 2)" and selecting "Save Mixture and Cluster Formulas."
Episode 8 (Original broadcast 4/23, rebroadcast 4/25) Useful functions (Tips) recommended for JMP beginners
In the final installment, we looked at how to use toolbars to make your work more efficient.
The toolbar is the area enclosed in a red frame in the figure below. By clicking an icon on the toolbar, you can launch the platform or operate tools with a single click.
There are many examples where the toolbar can help you work faster. For example, the Report window displays the following icons. From left to right, these are the Local Data Filter, Column Switcher, and Save Script to Data Table icons. These allow you to filter, switch columns, and save as a script with a single click.
● Display frequently used platforms on the toolbar
When drawing an Xbar control chart in JMP, you need to navigate through a deep hierarchy as shown below.
Rather than performing these operations repeatedly, you can add them to your toolbar so you can launch them with a single click.
Select View > Customize > Menus and Toolbars . The JMP Menu Editor opens, where you can do the following:
Create a new toolbar by right-clicking on the category you want to display on the toolbar and selecting Insert Before or Insert After.
In the right pane:
Caption : Enter the text that will be displayed when you move the cursor over the icon (in this example, Xbar Control Chart).
Built-in Command : Select the function you want to link to this icon. (In this example, ANALYZE:QUALITY AND PROCESS:XBAR CONTROL CHART)
Built-in Icons : Select the icons you want to display on the toolbar.
Once set, the icon you set will be displayed on the toolbar. (In this example, we have checked "Analysis" from "View" > "Toolbars."
Note: During the seminar, a question was asked in the chat section about whether there was a table of the English names of built-in commands and their corresponding menus, but unfortunately there is no such table. In that case, if you set JMP to English display*, the menu bar will be displayed in English, so the English names of those will correspond to the English names of the built-in commands.
Below is a table listing some platforms and their corresponding built-in commands.
| Platform Name | ビルトインコマンド |
| Analysis > Predictive Models > Partition | ANALYZE: PREDICTIVE MODELING PARTITION |
| Analyze > Advanced Modeling > Curve Fitting | ANALYZE: SPECIALIZED MODELING:NONLINEAR |
| Analyze > Multivariate > Multivariate Correlations | ANALYZE: MULTIVARIATE METHODS:MULTIVARIATE |
| Analyze > Multivariate > Principal Components Analysis | ANALYZE: MULTIVARIATE METHODS:PRINCIPAL COMPONENTS |
| Analyze > Cluster Analysis > Hierarchical Cluster Analysis | ANALYZE: CLUSTERING:HERACHIAL CLUSTER |
| Analyze > Reliability/Survival > Univariate Life | ANALYZE:RELIABILITY AND SURVIVAL:LIFE DISTRIBUTION |
| Analyze > Reliability/Survival > Survival | ANALYZE:RELIABILITY AND SURVIVAL:SURVIVAL |
*To display in English, select File > Preferences and change the display language to Japanese (Japanese) from the "Windows only" category.
by Naohiro Masukawa (JMP Japan)
This post originally written in Japanese and has been translated for your convenience. When you reply, it will also be translated back to Japanese.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.