/
DESCRIPTION
47 lines (47 loc) · 2.52 KB
/
DESCRIPTION
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
Package: learnerSelectoR
Type: Package
Title: An R Package for Automatically Selecting and Applying a (Machine) Learning Method
Version: 0.8.7
Author: Dr. Thomas Weise <tweise@hfuu.edu.cn>
Maintainer: Dr. Thomas Weise <tweise@hfuu.edu.cn>
Description: This package provides an infrastructure for selecting the right Machine Learning
method to process a given dataset automatically and then also applies it. This package thus
is something which will likely be used under-the-hood by our other packages to come. The goal
is to make a better, more robust choice regarding which Machine Learning method to use than
just applying all to the complete data and picking the best results. However, we here are
still at a very early stage in this development.
This package is intented to work together with other packages for model fitting or regression,
classification, and clustering. The goal is to provide an abstract method to select the right
model or learning algorithm: The main method, \code{\link{learning.learn}} is provided with
the input data and the size of the input data. It does not need to know the exact nature of
the data, though. It is further provided with functions that can be used to partition the
data based on a selection of sample indices. This way, it can internally realize
cross-validation. The method also is provided a set of learning methods (or models) in form
of a list of functions. After choosing the learning method or model which generalizes
better than the others and produces models of small size, this method is then applied to the
complete dataset and its result is returned. The learning method can also receive the input data
in different representations: Let's say you want to fit a model to some x-y data. Then you can
do that both on the original data as well as on log-scaled data. (Testing is always done on the
original data). This way you have two possible realizations of the same model. The one which
generalizes better is chosen.
License: LGPL-3
Encoding: UTF-8
LazyData: true
URL: http://www.github.com/thomasWeise/learnerSelectoR
BugReports: http://www.github.com/thomasWeise/learnerSelectoR/issues
Depends: R (>= 3.3.0)
Imports: methods (>= 3.3.2),
utilizeR (>= 0.8.4)
Suggests: testthat (>= 2.0.0),
roxygen2 (>= 6.0.1),
minpack.lm (>= 1.2.1)
Remotes: github::thomasWeise/utilizeR
RoxygenNote: 6.0.1.9000
Collate:
'check.R'
'Result.R'
'learn_compare.R'
'learn_check.R'
'rank.R'
'sample.R'
'learn.R'