One-Click Login with Social Account

Create A New Request

List Request Titles

 

Votes Comments Request
Resolved
asked 4 months ago by datamoose
updated 2 months ago by scottmac99
♥ 2

Rattle currently returns overall accuracy rate in the "evaluation" tab with the error matrix option selected. This is the total number of errors divided by the total number of data points (in the rele...
1
comment
asked 4 months ago by datamoose
updated 2 weeks ago by marjolein
♥ 2

Reason:   Jerome Friedman's "Rulefit"  is a simple, elegant and natural extension to gradient boosting, also remarkable for being a method that performs automatic feature selection.  In essence it...
7
comments
asked 4 months ago by datamoose
updated 3 months ago by sachinruk

Reason : Python has one. It makes things super fast on a desktop machine, particularly Deep Learning. Spec: A GPU arithmetic/algebra/whatever else is necessary  for R. Either a wrapper around an exis...
0
comments
asked 4 months ago by datamoose
updated 2 months ago by datamoose

randomForest is a suprisingly  limited package. There are many problems, including the inability to handle missing values and the limitation of 32 categories. Also there is the lack of a sensible cas...
2
comments
asked 3 months ago by khakieconomist
updated 3 months ago by khakieconomist
♥ 1

Typically, trees partition data recursively in order to maximise the information/purity/homogeneity of a single variable in the daughter nodes. It would be useful to have a tree that partitions the da...
1 2 3 8

One aspect of our mission involves better understanding current user preferences, usage patterns and desires. Therefore, periodically we will be asking our users to participate in short surveys that we hope will provide vital insights and analysis into what the R community is currently doing and how they want R to  develop and evolve.
After each survey, (conditional on having an adequate sample size) we will analyze and publish the results for everyone’s use. This is our first survey so please take a few mins to help out.



Connect R has been developed by long term R users, developers, and R user group leaders, who want to see R take its rightful place as the most commercially relevant data analysis tool. We plan on doing this by helping both R developers and users identify the developments that the community wants and needs. We want to make R the most responsively developed tool of its kind.

We also empower those many users who have neither the time nor funds to be R developers, to nevertheless get the R development they need. We in no way hamper, limit or deter the passion of existing developers, we simply give existing and new developers a view of what the huge and growing base of users wants from R.

Learn more about how we are adding a demand feedback loop to the already impressive R ecosystem here.