There are some articles that describe the issues with this data set. While it was very useful for illustrating predictive and exploratory modeling methods, it is time to put this example aside. It also highlights the need to find and understand the source of the data you are using when doing demonstrations and preparing documentation.
https://medium.com/@docintangible/racist-data-destruction-113e3eff54a8
https://scikit-learn.org/1.0/modules/generated/sklearn.datasets.load_boston.html
We will be continuing to look at our sample data library every release as a function of our normal release planning process. There is a process in place now for deprecating sample data, which is described on this Confluence page:
https://jmpdiscovery.atlassian.net/wiki/spaces/JMPDOC/pages/380633253/Process+for+Deprecating+Sample+Data
And in addition to the Boston Housing data, several other insensitive data sets have been removed or adjusted in JMP 18:
https://jmpdiscovery.atlassian.net/wiki/spaces/JMPDOC/pages/332890729/Deprecating+and+Changing+Sample+Data
Two other data sets that can be used for demonstrating regression and predicting modeling are the "Wine Quality" and "Abalone" data sets from the UCI Machine Learning Repository. These have good stories and demonstrate the modeling capabilities in JMP well:
https://archive.ics.uci.edu/dataset/1/abalone
https://archive.ics.uci.edu/dataset/186/wine+quality
Starting at JMP 18, the Boston Housing sample data table has been removed from the JMP samples. It was identified as containing problematic information – specifically, some of the data is discriminatory, unethical, and unfair.
“The Boston Housing data set has been removed from the Sample Data Folder and the end user documentation,” Product Management Director Dan Valente (@Daniel_Valente) said. “We should also stop using this data for demonstration to users and prospects.”
Python’s scikit-learn has already removed the data set and strongly discourages the use of the data unless the purpose of the code is to study and educate about ethical issues in data science and machine learning. “We intend to stay in step with any technology that JMP integrates with,” Technical Writing Manager Qiana Eaglin (@Qiana_Eaglin) said.
The following data sets have been identified as alternatives for regression examples: Diabetes, Liver Cancer, Body Fat, Fishing, and VA Lung Cancer. The Diabetes data set is used in the end-user documentation.
Changes to Sample Data have also been communicated to our EA customers via the Early Adopter page and will also be communicated externally via JMP Note.
Update (as of 12/05/23):
The following table summarizes the sample data that have been removed from JMP 18.0:
Sample Data
|
Reason for the change
|
Alternative tables that can be used
|
|---|
Crime.jmp
|
The table contains data that can be triggering, particularly the use of the term “rape”.
|
State Capitals.jmp, SAT.jmp
|
CrimeData.jmp
|
The table contains data that can be triggering, particularly the use of the term “rape”.
|
SATByYear.jmp, PopAgeGroup.jmp
|
Reactor Augment Design.jmp
|
The table is out of date, given changes to random numbers in the DOE design generation.
|
Hyde Miller is working on creating another augmentation file.
|
DotPlot.jsl
|
This script was developed in response to teachers' demands for dot plots. Graph Build now supports dot plots, making this script obsolete.
|
n/a
|
