I tend (being a bear of very small brain) to follow Frank Harrell on this https://www.fharrell.com/ given that I use software he's had input into most days.
In particular, the post "Road Map for Choosing Between Statistical Modeling and Machine Learning" https://www.fharrell.com/post/stat-ml/ and there's a follow-up (by Drew Levy) https://www.fharrell.com/post/stat-ml2/.
Apologies for the US spelling...
Most of the time I'm rarely interested in prediction, but rather in testing whether something has made a difference, which is rather different from the machine learning/AI advocates.
There's a lot of other thought-provoking posts there too.
Paul
Paul Bivand | Associate Director, Statistics & Analysis | Learning and Work Institute Paul Bivand | Cyfarwyddwr Cynorthwyol Ystadegau a Dadansoddiad |Sefydliad Dysgu a Gwaith
e. [log in to unmask] <mailto:[log in to unmask]> |<mailto:[log in to unmask] <mailto:[log in to unmask]> > | tw. @LWPaulBivand | dt. 020 7840 8335
w. www.learningandwork.org.uk <http://www.learningandwork.org.uk/> <http://www.learningandwork.org.uk/ <http://www.learningandwork.org.uk/> > | @LearnWorkUK | t. 020 7582 7221
3rd Floor | 89 Albert Embankment | London | SE1 7TP
https://www.learningandwork.org.uk/ourevents/ <https://www.learningandwork.org.uk/ourevents/>
-----Original Message-----
From: email list for Radical Statistics <[log in to unmask]> On Behalf Of Hennig, Christian
Sent: 03 March 2020 14:14
To: [log in to unmask]
Subject: Fw: Do statisticians hate algorithms?
I'm a statistician and my main methodological focus is cluster analysis. My impression is that cluster analysis is applied quite a lot, and that's a fairly "hot topic", although computer science maybe at least as strongly represented in the literature as statistics. There is a strong community working on "model-based clustering", mainly mixture models.
One reason why certain statisticians may not like cluster analysis and may not apply it that often is that the clustering problem is ambiguous, often not well defined, many clusterings can be obtained on the same data set with too little clear guideline how to decide between them, and on many data sets clustering is quite unstable.
When I was younger I contributed to a never ending stream of new cluster analysis approaches and methods (more than 100 appear every year), but now I think that what the area really needs more of is validation and decision support.
Not sure whether anyone's interested, but see, e.g., here:
https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Farxiv.org%2Fabs%2F1503.02059&data=02%7C01%7Cpaul.bivand%40LEARNINGANDWORK.ORG.UK%7C50826286ca7e4d23b64308d7bf7d2755%7C008187e87d444849a332a9fd32d2032d%7C1%7C0%7C637188416564151794&sdata=c0YCSyDFOWsGbTWEMD3sfAFnI6ULf8uXqE22WLU2Fx0%3D&reserved=0
...and here:
https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.sciencedirect.com%2Fscience%2Farticle%2Fpii%2FS0167865515001269&data=02%7C01%7Cpaul.bivand%40LEARNINGANDWORK.ORG.UK%7C50826286ca7e4d23b64308d7bf7d2755%7C008187e87d444849a332a9fd32d2032d%7C1%7C0%7C637188416564151794&sdata=I0yxqyjyuBLex9AWbIqajAsgbMhJgYc1Em82IHGXh7U%3D&reserved=0
Best wishes,
Christian
> David Byrne write:
Interesting sub-text at last week's excellent conference when people started talking about big data - it seems that statisticians don't like methods based on algorithms. Now cluster analysis which has been around for more than 40 years is algorithm based, albeit that the mathematical basis of it is accessible in a way that the coding schemes of learning algorithms are not, and has in my view been far too little used in exploratory data analysis. Is that also down to a prejudice against algorithms?
David Byrne Ph.D., FAcSS
see my recent book:
Class after Industry: a complex realist approach
https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.palgrave.com%2Fgb%2Fbook%2F9783030026431&data=02%7C01%7Cpaul.bivand%40LEARNINGANDWORK.ORG.UK%7C50826286ca7e4d23b64308d7bf7d2755%7C008187e87d444849a332a9fd32d2032d%7C1%7C0%7C637188416564151794&sdata=jDzaSSga25F9SVSfVi2Ew94APFbycX1NXk7b2I%2Bt1bg%3D&reserved=0
****************************************************** Please note that if you press the 'Reply' button your message will go only to the sender of this message. If you want to reply to the whole list, use your mailer's 'Reply-to-All' button to send your message automatically to [log in to unmask]<mailto:[log in to unmask]>. Disclaimer: The messages sent to this list are the views of the sender and cannot be assumed to be representative of the range of views held by subscribers to the Radical Statistics Group. To find out more about Radical Statistics and its aims and activities and read current and past issues of our newsletter you are invited to visit our web site https://eur02.safelinks.protection.outlook.com/?url=www.radstats.org.uk&data=02%7C01%7Cpaul.bivand%40LEARNINGANDWORK.ORG.UK%7C50826286ca7e4d23b64308d7bf7d2755%7C008187e87d444849a332a9fd32d2032d%7C1%7C0%7C637188416564151794&sdata=KU54Dx05AqcnpN%2FEA%2FUdYT4BBYvM1oQVMXB%2ByWz%2B2JU%3D&reserved=0<https://eur02.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.radstats.org.uk&data=02%7C01%7Cpaul.bivand%40LEARNINGANDWORK.ORG.UK%7C50826286ca7e4d23b64308d7bf7d2755%7C008187e87d444849a332a9fd32d2032d%7C1%7C0%7C637188416564151794&sdata=sTqeZXxzlMIeU3QUWh29TciC8CuwSOMu6V3fo%2B1xia4%3D&reserved=0>. *******************************************************
******************************************************
Please note that if you press the 'Reply' button your message will go only to the sender of this message.
If you want to reply to the whole list, use your mailer's 'Reply-to-All' button to send your message automatically to [log in to unmask]
Disclaimer: The messages sent to this list are the views of the sender and cannot be assumed to be representative of the range of views held by subscribers to the Radical Statistics Group. To find out more about Radical Statistics and its aims and activities and read current and past issues of our newsletter you are invited to visit our web site https://eur02.safelinks.protection.outlook.com/?url=www.radstats.org.uk&data=02%7C01%7Cpaul.bivand%40LEARNINGANDWORK.ORG.UK%7C50826286ca7e4d23b64308d7bf7d2755%7C008187e87d444849a332a9fd32d2032d%7C1%7C0%7C637188416564151794&sdata=KU54Dx05AqcnpN%2FEA%2FUdYT4BBYvM1oQVMXB%2ByWz%2B2JU%3D&reserved=0.
*******************************************************
This message contains confidential information and is intended only for the individual(s) addressed in the message. If you are not the named addressee, you should not disseminate, distribute, or copy this e-mail. If you are not the intended recipient, you are notified that disclosing, distributing, or copying this e-mail is strictly prohibited.
******************************************************
Please note that if you press the 'Reply' button your
message will go only to the sender of this message.
If you want to reply to the whole list, use your mailer's
'Reply-to-All' button to send your message automatically
to [log in to unmask]
Disclaimer: The messages sent to this list are the views of the sender and cannot be assumed to be representative of the range of views held by subscribers to the Radical Statistics Group. To find out more about Radical Statistics and its aims and activities and read current and past issues of our newsletter you are invited to visit our web site www.radstats.org.uk.
*******************************************************
|