The JMP Japan Division is holding a seminar called "JMP Information Bureau" every Tuesday for a total of eight sessions from March to April of this year. The purpose of these seminars is to convey a variety of information about JMP in a short 30-minute session, and the number of participants exceeds expectations.
JMP Information Bureau opening
Purpose of “JMP Information Bureau”
The subtitle of this seminar is "Making JMP easier to understand and more fun." We aim to hold seminars with the following points in mind:
- This is an opportunity to learn about the various functions and information of JMP.
- An opportunity to learn about the fun of data analysis and statistical analysis
Additionally, we provide information on JMP and statistics that is rarely or impossible to touch upon in regular seminars.
JMP Information Bureau Details and application page (accepted until April 25th)
JMP Information Bureau ~Make JMP more understandable and fun~ | JMP
At the time of writing this blog, the fourth seminar has already ended. Therefore, I have summarized the main contents that have been covered so far. If you have attended the seminar, I think that by reading this article as a review, you will be able to better retain what you have learned.
*The main contents of episodes 5 to 8 (final episode) will be posted on a separate blog after the 8th episode is over.
Part 1 (main broadcast 3/5, rebroadcast 3/7) Useful functions (Tips) recommended for JMP beginners
We explained the JMP options [Restart analysis] and [Automatic recalculation] .
You can redo it as many times as you like - Learn how to use it by "restarting analysis"
JMP includes a variety of sample data, many of which include scripts for analysis.
For example, if you look at the JMP sample data "Car Poll.jmp", you will see that various scripts are saved. If I want to perform "multiple correspondence analysis using additional variables", how should I operate it? Run the appropriate script and try the analysis again.
From the red triangle at the top left of the report, select Redo > Restart Analysis .
Then, a window for specifying columns will appear, and you can see that "Country of origin" and "Size" have been specified as [Y, objective variable], and "Gender" has been specified as [Z, additional variable].
What happens to the analysis report if the data changes? ～How “automatic recalculation” works～
In JMP, you can draw mosaic diagrams using either the [Bivariate Relationship] or [Graph Builder] platforms. For example, after drawing a mosaic diagram using the sample data "Car Poll.jmp" from earlier, we return to the data table and show the report after excluding the first 20 rows.
When excluding 20 rows, [Graph Builder] redraws the mosaic diagram excluding 20 rows of data. On the other hand, the mosaic diagram is not redrawn in [Bivariate relationship]. This difference depends on whether [Automatic recalculation] is enabled or not.
Platforms such as Graph Builder under the Graph menu have Auto Recalc turned on by default. So if you exclude or change data, these changes will be reflected in the graph. However, many platforms, such as Bivariate Relationships under the Analysis menu, have Auto Recalculation turned off by default. If the data changes, the changes are not reflected. If you want the changes to be reflected in the graph, you must manually turn on Auto Recalc.
Part 2 (main broadcast 3/12, rebroadcast 3/14) Learn statistics interactively using JMP
to JMPis equipped with many sample scripts that allow you to learn statistical concepts interactively. In this episode, we introduced scripts in demo format that will help you understand regression analysis, correlation, confidence intervals, and central limit theorem.
JMP sample data and scripts introduced at the seminar
●Correlation, regression
- demoCorr.jsl
- Corrsim.jmp
- demoLeastSquares.jsl
In "demoLeastSquares.jsl", you can learn the concept of least squares method by drawing a straight line to fit the plot points and checking the resulting residual sum of squares.
Confidence interval
"Condidence.jsl" allows you to learn the concept of confidence intervals through random number simulation.
A 95% confidence interval for a mean is an interval within which there is an estimated 95% probability that the true mean of a data set lies within that interval.
The figure below shows the average5, standard deviation1from the distribution of20samples were extracted to construct a 95% confidence interval and to examine whether they contained the true population mean of 5. This process was repeated 100 times to observe whether the confidence intervals contained the true mean value of 5. Black bars indicate confidence intervals that contained 5, and red bars indicate those that did not. In this example, 95 out of 100 confidence intervals contained the population mean of 5. This is the concept of a confidence interval.
●Central limit theorem
- Sampling Distribution of Sample Means.jsl
In "Sampling Distribution of Sample Means.jsl", you can learn the "Central Limit Theorem", which is an important theorem in statistics.
The central limit theorem is roughly stated as follows:
No matter what the distribution of the population is, the sample mean can be approximated by a normal distribution.
In the example below, 25 random numbers are extracted from a uniform distribution, and their distribution is displayed as a histogram in the "Distribution of Sample Data" section in the middle. The average of these 25 numbers is the sample mean. By repeatedly extracting random numbers about 1,800 times, 1,800 sample means are obtained. The "Distribution of Sample Means" section in the bottom section visualizes the distribution of these 1,800 sample means as a histogram.
As you can see from the histogram of "Distribution of Sample Means", you can visually confirm that the distribution of sample means follows a normal distribution.
Episode 3 (main broadcast 3/19, rebroadcast 3/21) About Excel profiles
We demonstrated the functionality of "Excel Profile", which allows you to visualize models saved in Microsoft Excel using JMP profiles.
- By using JMP profiles, you can see how multiple input variables affect output variables in formulas (models) set in Excel.
- Profiles have a feature called "Simulator". Assuming a distribution for X, we can simulate the distribution of Y using Monte Carlo simulation.
Relatedly, I am writing a blog about how to complete the equivalent of an Excel profile using just JMP.
How to do everything that can be done with "Excel profile" using just JMP profile - JMP User Community
Episode 4 (Original broadcast 3/26, rebroadcast 3/28) Taylor Swift song analysis
We analyzed the words that are often included in the lyrics of the songs included in Taylor Swift's albums (148 songs) and the factors that make them popular.
Through this analysis, we have shown important points when analyzing data.
- Learn and investigate the contents of the data in detail.
- Consider and interpret analysis results by making full use of unique technology and knowledge.
Below is a word cloud of the lyrics.
by Naohiro Masukawa (JMP Japan)
Naohiro Masukawa - JMP User Community
