Difference between revisions of "Orange: AdaBoost"

From OnnoWiki
Jump to navigation Jump to search
 
(5 intermediate revisions by the same user not shown)
Line 3: Line 3:
  
  
An ensemble meta-algorithm that combines weak learners and adapts to the ‘hardness’ of each training sample.
+
Widget AdaBoost adalah sebuah ensemble meta-algoritma yang menggabungkan learner yang lemah dan beradaptasi dengan 'kekerasan' masing-masing sampel training.
  
Inputs
+
==Input==
  
    Data: input dataset
+
Data: input dataset
    Preprocessor: preprocessing method(s)
+
Preprocessor: preprocessing method(s)
    Learner: learning algorithm
+
Learner: learning algorithm
  
Outputs
+
==Output==
  
    Learner: AdaBoost learning algorithm
+
Learner: AdaBoost learning algorithm
    Model: trained model
+
Model: trained model
  
The AdaBoost (short for “Adaptive boosting”) widget is a machine-learning algorithm, formulated by Yoav Freund and Robert Schapire. It can be used with other learning algorithms to boost their performance. It does so by tweaking the weak learners.
+
Widget AdaBoost (singkatan dari “Adaptive boosting”) adalah algoritma machine-learning, yang di formulasikan oleh Yoav Freund and Robert Schapire. Ini dapat digunakan dengan algoritma learner lainnya untuk meningkatkan kinerja mereka. Itu dilakukan dengan mengutak-atik learner yang lemah.
  
AdaBoost works for both classification and regression.
+
Widget AdaBoost dapat bekerja baik untuk task classification maupun regression.
  
 
[[File:AdaBoost-stamped.png|center|200px|thumb]]
 
[[File:AdaBoost-stamped.png|center|200px|thumb]]
  
    The learner can be given a name under which it will appear in other widgets. The default name is “AdaBoost”.
+
* The learner can be given a name under which it will appear in other widgets. The default name is “AdaBoost”.
    Set the parameters. The base estimator is a tree and you can set:
+
* Set the parameters. The base estimator is a tree and you can set:
 +
** Number of estimators
 +
** Learning rate: it determines to what extent the newly acquired information will override the old information (0 = the agent will not learn anything, 1 = the agent considers only the most recent information)
 +
** Fixed seed for random generator: set a fixed seed to enable reproducing the results.
  
        Number of estimators
+
* Boosting method.
        Learning rate: it determines to what extent the newly acquired information will override the old information (0 = the agent will not learn anything, 1 = the agent considers only the most recent information)
+
** Classification algorithm (if classification on input): SAMME (updates base estimator’s weights with classification results) or SAMME.R (updates base estimator’s weight with probability estimates).
        Fixed seed for random generator: set a fixed seed to enable reproducing the results.
+
** Regression loss function (if regression on input): Linear (), Square (), Exponential ().
  
    Boosting method.
+
* Produce a report.
        Classification algorithm (if classification on input): SAMME (updates base estimator’s weights with classification results) or SAMME.R (updates base estimator’s weight with probability estimates).
+
* Click Apply after changing the settings. That will put the new learner in the output and, if the training examples are given, construct a new model and output it as well. To communicate changes automatically tick Apply Automatically.
        Regression loss function (if regression on input): Linear (), Square (), Exponential ().
 
    Produce a report.
 
 
 
    Click Apply after changing the settings. That will put the new learner in the output and, if the training examples are given, construct a new model and output it as well. To communicate changes automatically tick Apply Automatically.
 
  
 
==Contoh==
 
==Contoh==
  
For classification, we loaded the iris dataset. We used AdaBoost, Tree and Logistic Regression and evaluated the models’ performance in Test & Score.
+
Untuk classification, kita load iris dataset. Kita menggunakan widget AdaBoost, widget Tree dan widget Logistic Regression dan mengevaluasi performance model di widget Test & Score.
 
 
[[File:AdaBoost-classification.png|center|200px|thumb]]
 
 
 
For regression, we loaded the housing dataset, sent the data instances to two different models (AdaBoost and Tree) and output them to the Predictions widget.
 
 
 
[[File:AdaBoost-regression.png|center|200px|thumb]]
 
 
 
  
 +
[[File:AdaBoost-classification.png|center|600px|thumb]]
  
 +
Untuk regression, kita load housing dataset, kirim data instances ke dua model yang berbeda (widget AdaBoost dan widget Tree) dan keluarannya ke widget Predictions.
  
 +
[[File:AdaBoost-regression.png|center|600px|thumb]]
  
 
==Referensi==
 
==Referensi==

Latest revision as of 10:14, 6 April 2020

Sumber: https://docs.biolab.si//3/visual-programming/widgets/model/adaboost.html


Widget AdaBoost adalah sebuah ensemble meta-algoritma yang menggabungkan learner yang lemah dan beradaptasi dengan 'kekerasan' masing-masing sampel training.

Input

Data: input dataset
Preprocessor: preprocessing method(s)
Learner: learning algorithm

Output

Learner: AdaBoost learning algorithm
Model: trained model

Widget AdaBoost (singkatan dari “Adaptive boosting”) adalah algoritma machine-learning, yang di formulasikan oleh Yoav Freund and Robert Schapire. Ini dapat digunakan dengan algoritma learner lainnya untuk meningkatkan kinerja mereka. Itu dilakukan dengan mengutak-atik learner yang lemah.

Widget AdaBoost dapat bekerja baik untuk task classification maupun regression.

AdaBoost-stamped.png
  • The learner can be given a name under which it will appear in other widgets. The default name is “AdaBoost”.
  • Set the parameters. The base estimator is a tree and you can set:
    • Number of estimators
    • Learning rate: it determines to what extent the newly acquired information will override the old information (0 = the agent will not learn anything, 1 = the agent considers only the most recent information)
    • Fixed seed for random generator: set a fixed seed to enable reproducing the results.
  • Boosting method.
    • Classification algorithm (if classification on input): SAMME (updates base estimator’s weights with classification results) or SAMME.R (updates base estimator’s weight with probability estimates).
    • Regression loss function (if regression on input): Linear (), Square (), Exponential ().
  • Produce a report.
  • Click Apply after changing the settings. That will put the new learner in the output and, if the training examples are given, construct a new model and output it as well. To communicate changes automatically tick Apply Automatically.

Contoh

Untuk classification, kita load iris dataset. Kita menggunakan widget AdaBoost, widget Tree dan widget Logistic Regression dan mengevaluasi performance model di widget Test & Score.

AdaBoost-classification.png

Untuk regression, kita load housing dataset, kirim data instances ke dua model yang berbeda (widget AdaBoost dan widget Tree) dan keluarannya ke widget Predictions.

AdaBoost-regression.png

Referensi

Pranala Menarik