Hi ftr,
Very interesting! I've never heard of ViSta or Tanagra, but I'll definitely take a look at them. As you can see from my regularly-updated web page, I'm just slightly obsessed with data science software: http://r4stats.com/articles/popularity/.
In the open-source data mining realm, KNIME has much to recommend it. It hooks into many other open source tools. However, it's particularly weak in statistical analysis. I sent them a list of over 40 tests it lacks & those are on their future roadmap, but at the rate they're going, it'll take years to catch up with SPSS.
I completely agree with you on the importance of output management. That's one of the topics that clearly separates BlueSky from jamovi. BlueSky uses R's "Tidyverse" and "broom" functions to do things like perform regression by a categorical variable, then output 3 different data sets, one at the model level (e.g. r**2) one at the parameter level (betas, p-values, etc) and one at the observation level (predicted value, measures of influence, etc.) All its modeling methods, statistical and machine learning, can be used to make predictions on the same or new data sets.
It has been a while since I tried Mondrian. As I recall, it did a beautiful job on categorical visualization, but little else way back when I tried it last. I'll take another look.
SPSS & SAS offer modern methods for imputing missing values, but when it comes to point-and-click competition, support is weak across the board. I pointed this out to the BlueSky developers & they say they'll add support for R's Simputation package in their January release: https://cran.r-project.org/web/packages/simputation/vignettes/intro.html . Simputation acts as front-end to many other R packages, so that should provide a wide range of options.
I have to confess that I didn't look into categorical methods very thoroughly. I'll see if I can't add that to my review template (missing value too).
Cheers,
Bob
-----Original Message-----
From: ftr <***@free.fr>
Sent: Thursday, November 29, 2018 9:50 AM
To: Muenchen, Robert A (Bob) <***@utk.edu>; pspp-***@gnu.org
Subject: Re: Other Free SPSS Alternatives
Hi Bob,
I really appreciate your work as I am not a programmer but a social scientist that uses stats to answer substantial questions. And I was just "surprised" when I asked a newbie question in the R forum some years ago. So SPSS for SPPS' sake, R for R's sake, is not my job.
Therefore I bought your book, too.
There are two other free programs that merit attention:
ViSta : https://en.wikipedia.org/wiki/ViSta,_The_Visual_Statistics_system
Mondrian: http://www.theusrus.de/Mondrian/ Mondrian is a R GUI.
Tanagra: http://eric.univ-lyon2.fr/~ricco/tanagra/en/tanagra.html
TANAGRA is a free DATA MINING software for academic and research purposes. It proposes several data mining methods from exploratory data analysis, statistical learning, machine learning and databases area.
The advantage of Tanagra ist that its programmer as a data mining teacher and as such he published a manuals for a large number of his sub-programmes.
As for commercial side I don't see anything on the horizon for Jasp and jamovi as they are both academic programmes.
Both have a real drawback: You can't create new variables from the results of the procedures. So you get rapid and good looking results from a PCA, for instance, but you can't save the dimensions. I asked for this but it does not seem on their radar. This makes these programmes less useful as it breaks the analysis and production process. They solved the first problem of an analysis cycle, data input , but not the last one, data output.
BTW, two other issues I would treat in a programme review is:
how easily treated are missing values ?
Missing values are a standard problem in real life, and already input with MV varies from programme to programme. Does the programme do more than list-wise or case-wise deletion, or does it already stall when missing values appear ?
Is the program useful for categorical data analysis ? Very often the statistical programmes are oriented towards the analysis of continuous data (in biomedical, physics, engineering, ...) , but in social, political or management sciences or in any survey analysis this is not at the centre. For instance, multiple correspondence analysis is quite seldom.
Regards,
ftr
Post by Muenchen, Robert A (Bob)Hi ftr,
I'm aware of Jasp, but haven't decided if I'll review it yet. I still haven't gotten the jamovi review into the extensive review format that I used to standardize the other reviews. I hope to do that over the Christmas break. I'd also like to look at R AnalyticFlow, which is more like SPSS Modeler than SPSS Statistics: https://r.analyticflow.com/en/. That workflow interface has much to recommend it, though reporting seems to be a weak point. I love the APA-style tables in jamovi and BlueSky!
Jamovi was started by former Jasp developers. There's a very interesting interview with Jonathan Love about those two packages here: http://blog.efpsa.org/2017/03/23/introducing-jamovi-free-and-open-statistical-software-combining-ease-of-use-with-the-power-of-r/ .
I'm using the open source version of BlueSky, but I'll be surprised if there aren't commercial angles to several of the others for people who want to be able to pay for tech support.
With so many wonderful options, it's a great time to be using statistics &/or machine learning. I think IBM/SPSS, SAS Institute, and Statacorp must be quaking in their boots!
Cheer,
Bob
-----Original Message-----
Sent: Wednesday, November 28, 2018 6:23 PM
Subject: Re: Other Free SPSS Alternatives
Hi Bob,
Did you wrote about Jasp ?
Blue Sky Stats has an open source and a commercial version.
Regards,
-ftr
Post by Muenchen, Robert A (Bob)Hi All,
I've been looking into open-source SPSS alternatives and have written
http://r4stats.com/articles/software-reviews/
Several of them are adding features at a fast rate, so this should be
very interesting area to watch.
http://r4stats.com/2018/10/01/bluesky-statistics-6-04-gui-for-r-updat
e
/
Cheers,
Bob
*Bob Muenchen*
Manager, Research Computing Support
The University of Tennessee, Knoxville Office of Information
Technology
522 Greve Hall
821 Volunteer Blvd.
Knoxville, TN 37996-3395
865-974-5230
http://oit.utk.edu/research
*OIT RESEARCH COMPUTING SUPPORT*
Statistical Consulting . Graphics & Visualization . Web Survey Design
. Qualitative Analysis . Mapping & GIS Stay informed. S
<http://utk.us3.list-manage.com/subscribe?u=a348841b15d00f4d14e00fb59
& id=8a0f6d9e0d>ign up to receive monthly instructor and researcher
updates
<http://utk.us3.list-manage.com/subscribe?u=a348841b15d00f4d14e00fb59&id=8a0f6d9e0d>.
_______________________________________________
Pspp-users mailing list
https://lists.gnu.org/mailman/listinfo/pspp-users
---
L'absence de virus dans ce courrier électronique a été vérifiée par le logiciel antivirus Avast.
https://www.avast.com/antivirus
_______________________________________________
Pspp-users mailing list
https://lists.gnu.org/mailman/listinfo/pspp-users