search

jasp-stats / jasp-issues

This repository is solely meant for reporting of bugs, feature requests and other issues in JASP.
58 stars 29 forks source link

Some features to improve JASP #1407

Open PetBuj opened 2 years ago

PetBuj commented 2 years ago
* Enhancement: 1. ANOVA --> using more than one dependent variable 2. Possibility to sort frequency (or any other) output table 3. Enhance Lin. regression by All possible regressions feature (to select the most proper covariates) 4. Non-hierarchical clustering --> compute variability measure from 2 up to a given number of clusters (it is very useful) 5. In hierarchical clustering --> possibility to print labels in the dendrogram 6. Add some known time-series methods * Purpose: All my six suggestions are (I hope) very useful, not only in the teaching statistics but also in some academic data analysis. * Use-case: All my six suggestions are also very useful, all are very factual (not theoretical) **Is your feature request related to a problem? Please describe.** All my suggestions were raised from my practice of JASP. The program is very useful and nice and a small count of (I hope) simple enhancements make it more practical. **Describe the solution you'd like** I am not a pure software developer, but regarding my suggestions, I hope, there is a lot of commercial or open statistical programs where the enhancements are employed. **Describe alternatives you've considered** N/A **Additional context**

N/A

vandenman commented 2 years ago

Hi @PetBuj Thanks for the suggestions!

  1. ANOVA --> using more than one dependent variable

You can do a MANOVA for this purpose. A MANOVA takes the correlation between the dependent variables into account whereas multiple independent ANOVAs ignore this.

  1. Possibility to sort frequency (or any other) output table

Good idea!

  1. Enhance Lin. regression by All possible regressions feature (to select the most proper covariates)

Bayesian linear regression does this. In a frequentist setting, this is a bit more tricky because it can blow up the type-I error rate.

  1. Non-hierarchical clustering --> compute variability measure from 2 up to a given number of clusters (it is very useful)

Could you elaborate a bit on this? Is there a reference or R package that implements this so I can take a look or forward it to other members of the team?

  1. In hierarchical clustering --> possibility to print labels in the dendrogram

Good idea!

  1. Add some known time-series methods

This is on our to-do list. Some time-series methods should be available in the upcoming release.

PetBuj commented 2 years ago

Hi Don,

Thank you for so fast response and also for clarification of some my remarks. Regarding #4, for example in NCSS, there is a table of the odd of withinn-clusters variability and total variability. Then, it is clear how the higer number of cluster decreases the ‚inner‘ variability of clusters. Maybe, there is somethig, but I didn’t find it.

Best regards,

P. Bujok

From: Don van den Bergh @.> Sent: Tuesday, September 7, 2021 9:17 AM To: jasp-stats/jasp-issues @.> Cc: Petr Bujok @.>; Mention @.> Subject: Re: [jasp-stats/jasp-issues] Some features to improve JASP (#1407)

Hi @PetBujhttps://github.com/PetBuj Thanks for the suggestions!

  1. ANOVA --> using more than one dependent variable

You can do a MANOVA for this purpose. A MANOVA takes the correlation between the dependent variables into account whereas multiple independent ANOVAs ignore this.

  1. Possibility to sort frequency (or any other) output table

Good idea!

  1. Enhance Lin. regression by All possible regressions feature (to select the most proper covariates)

Bayesian linear regression does this. In a frequentist setting, this is a bit more tricky because it can blow up the type-I error rate.

  1. Non-hierarchical clustering --> compute variability measure from 2 up to a given number of clusters (it is very useful)

Could you elaborate a bit on this? Is there a reference or R package that implements this so I can take a look or forward it to other members of the team?

  1. In hierarchical clustering --> possibility to print labels in the dendrogram

Good idea!

  1. Add some known time-series methods

This is on our to-do list. Some time-series methods should be available in the upcoming release.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/jasp-stats/jasp-issues/issues/1407#issuecomment-914053637, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AVQORQKJY7CEFK2YMDITBOLUAW347ANCNFSM5DRTLEUQ. Triage notifications on the go with GitHub Mobile for iOShttps://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Androidhttps://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

tomtomme commented 6 months ago

@PetBuj Time-Series modules have been implemented. But the following items not. I checked in jasp 0.18.3:

Maybe you can rename the issue to "non-hierarchical clustering improvements"