Orange: Stacking - Revision history

Onnowpurbo: /* Output */

2020-04-06T04:06:26Z

Output

← Older revision		Revision as of 04:06, 6 April 2020
Line 16:		Line 16:
	Model: trained model		Model: trained model

	~~Stacking is an ensemble method that computes a meta model from several base models. The Stacking widget has the Aggregate input,~~Stacking adalah metode ensemble yang menghitung model meta dari beberapa model dasar. Widget Stacking memiliki input agregat, yang menyediakan metode untuk menggabungkan model input. Jika tidak ada input agregasi yang diberikan, metode default akan digunakan. Model default tersebut adalah Logistic Regression untuk classification (klasifikasi) dan Ridge Regression untuk regression (regresi).		Stacking adalah metode ensemble yang menghitung model meta dari beberapa model dasar. Widget Stacking memiliki input agregat, yang menyediakan metode untuk menggabungkan model input. Jika tidak ada input agregasi yang diberikan, metode default akan digunakan. Model default tersebut adalah Logistic Regression untuk classification (klasifikasi) dan Ridge Regression untuk regression (regresi).

	[[File:Stacking-stamped.png\|center\|200px\|thumb]]		[[File:Stacking-stamped.png\|center\|200px\|thumb]]

Onnowpurbo at 11:28, 5 April 2020

2020-04-05T11:28:32Z

← Older revision		Revision as of 11:28, 5 April 2020
Line 2:		Line 2:


	~~Stack~~ (menumpuk) multiple model.		Widget Stacking berfungsi untuk stacking (menumpuk) multiple model.

	==Input==		==Input==

Onnowpurbo: /* Contoh */

2020-04-05T11:27:42Z

Contoh

← Older revision		Revision as of 11:27, 5 April 2020
Line 30:		Line 30:
	Mari kita gunakan stacking. Stacking membutuhkan beberapa learner pada input dan metode agregasi. Dalam kasus kita, ini adalah widget Logistic Regression. Sebuah constructed meta learner dikirim ke widget Test & Score. Hasil telah meningkat, walaupun hanya sedikit. Stacking biasanya bekerja dengan baik pada dataset yang kompleks.		Mari kita gunakan stacking. Stacking membutuhkan beberapa learner pada input dan metode agregasi. Dalam kasus kita, ini adalah widget Logistic Regression. Sebuah constructed meta learner dikirim ke widget Test & Score. Hasil telah meningkat, walaupun hanya sedikit. Stacking biasanya bekerja dengan baik pada dataset yang kompleks.

	[[File:Stacking-Example.png\|center\|~~200px~~\|thumb]]		[[File:Stacking-Example.png\|center\|600px\|thumb]]

	==Referensi==		==Referensi==

Onnowpurbo: /* Contoh */

2020-04-05T11:27:33Z

Contoh

← Older revision		Revision as of 11:27, 5 April 2020
Line 26:		Line 26:
	==Contoh==		==Contoh==

	~~Kami~~ akan menggunakan widget Paint Data untuk menunjukkan bagaimana widget digunakan. ~~Kami~~ paint dataset yang kompleks dengan 4 label class dan mengirimkannya ke widget Test & Score. Kita juga menyediakan tiga learner kNN, masing-masing dengan parameter yang berbeda (jumlah neighbour adalah 5, 10 atau 15). Hasil evaluasi yang dihasilkan baik, tetapi bisakah kita membuatnya lebih baik?		Kita akan menggunakan widget Paint Data untuk menunjukkan bagaimana widget Staking digunakan. Kita paint dataset yang kompleks dengan 4 label class dan mengirimkannya ke widget Test & Score. Kita juga menyediakan tiga learner widget kNN, masing-masing dengan parameter yang berbeda (jumlah neighbour adalah 5, 10 atau 15). Hasil evaluasi yang dihasilkan baik, tetapi bisakah kita membuatnya lebih baik?

	Mari kita gunakan stacking. Stacking membutuhkan beberapa learner pada input dan metode agregasi. Dalam kasus kita, ini adalah Logistic Regression. Sebuah constructed meta learner dikirim ke widget Test & Score. Hasil telah meningkat, walaupun hanya sedikit. Stacking biasanya bekerja dengan baik pada dataset yang kompleks.		Mari kita gunakan stacking. Stacking membutuhkan beberapa learner pada input dan metode agregasi. Dalam kasus kita, ini adalah widget Logistic Regression. Sebuah constructed meta learner dikirim ke widget Test & Score. Hasil telah meningkat, walaupun hanya sedikit. Stacking biasanya bekerja dengan baik pada dataset yang kompleks.

	[[File:Stacking-Example.png\|center\|200px\|thumb]]		[[File:Stacking-Example.png\|center\|200px\|thumb]]



	==Referensi==		==Referensi==

Onnowpurbo at 21:52, 6 March 2020

2020-03-06T21:52:57Z

← Older revision		Revision as of 21:52, 6 March 2020
Line 2:		Line 2:


	Stack multiple ~~models~~.		Stack (menumpuk) multiple model.

	==Input==		==Input==
Line 16:		Line 16:
	Model: trained model		Model: trained model

	Stacking is an ensemble method that computes a meta model from several base models. The Stacking widget has the Aggregate input, ~~which provides a method for aggregating the~~ input ~~models~~. ~~If no aggregation~~ input ~~is given the~~ default ~~methods are used~~. ~~Those are~~ Logistic Regression ~~for~~ classification ~~and~~ Ridge Regression ~~for~~ regression ~~problems~~.		Stacking is an ensemble method that computes a meta model from several base models. The Stacking widget has the Aggregate input,Stacking adalah metode ensemble yang menghitung model meta dari beberapa model dasar. Widget Stacking memiliki input agregat, yang menyediakan metode untuk menggabungkan model input. Jika tidak ada input agregasi yang diberikan, metode default akan digunakan. Model default tersebut adalah Logistic Regression untuk classification (klasifikasi) dan Ridge Regression untuk regression (regresi).

	[[File:Stacking-stamped.png\|center\|200px\|thumb]]		[[File:Stacking-stamped.png\|center\|200px\|thumb]]
Line 26:		Line 26:
	==Contoh==		==Contoh==

	~~We will use~~ Paint Data ~~to demonstrate how the~~ widget ~~is used~~. ~~We painted a complex~~ dataset ~~with~~ 4 class ~~labels and sent it to~~ Test & Score. ~~We also provided three~~ kNN ~~learners~~, ~~each with a different parameters~~ (~~number of neighbors is~~ 5, 10 or 15). ~~Evaluation results are good~~, ~~but can we do better~~?		Kami akan menggunakan widget Paint Data untuk menunjukkan bagaimana widget digunakan. Kami paint dataset yang kompleks dengan 4 label class dan mengirimkannya ke widget Test & Score. Kita juga menyediakan tiga learner kNN, masing-masing dengan parameter yang berbeda (jumlah neighbour adalah 5, 10 atau 15). Hasil evaluasi yang dihasilkan baik, tetapi bisakah kita membuatnya lebih baik?

	~~Let’s use Stacking~~. Stacking ~~requires several learners on the~~ input ~~and an aggregation method~~. ~~In our case~~, ~~this is~~ Logistic Regression. A constructed meta learner ~~is then sent to~~ Test & Score. ~~Results have improved~~, ~~even if only marginally~~. Stacking ~~normally works well on complex data sets~~.		Mari kita gunakan stacking. Stacking membutuhkan beberapa learner pada input dan metode agregasi. Dalam kasus kita, ini adalah Logistic Regression. Sebuah constructed meta learner dikirim ke widget Test & Score. Hasil telah meningkat, walaupun hanya sedikit. Stacking biasanya bekerja dengan baik pada dataset yang kompleks.

	[[File:Stacking-Example.png\|center\|200px\|thumb]]		[[File:Stacking-Example.png\|center\|200px\|thumb]]

Onnowpurbo at 04:32, 28 January 2020

2020-01-28T04:32:25Z

← Older revision		Revision as of 04:32, 28 January 2020
Line 20:		Line 20:
	[[File:Stacking-stamped.png\|center\|200px\|thumb]]		[[File:Stacking-stamped.png\|center\|200px\|thumb]]

	The meta learner can be given a name under which it will appear in other widgets. The default name is “Stack”.		* The meta learner can be given a name under which it will appear in other widgets. The default name is “Stack”.
	Click Apply to commit the aggregated model. 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.		* Click Apply to commit the aggregated model. 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.
	Access help and produce a report.		* Access help and produce a report.

	==Contoh==		==Contoh==

Onnowpurbo at 04:26, 28 January 2020

2020-01-28T04:26:38Z

← Older revision		Revision as of 04:26, 28 January 2020
Line 4:		Line 4:
	Stack multiple models.		Stack multiple models.

	~~Inputs~~		==Input==

	Data: input dataset		Data: input dataset
	Preprocessor: preprocessing method(s)		Preprocessor: preprocessing method(s)
	Learners: learning algorithm		Learners: learning algorithm
	Aggregate: model aggregation method		Aggregate: model aggregation method

	~~Outputs~~		==Output==

	Learner: aggregated (stacked) learning algorithm		Learner: aggregated (stacked) learning algorithm
	Model: trained model		Model: trained model

Onnowpurbo at 03:25, 23 January 2020

2020-01-23T03:25:59Z

← Older revision		Revision as of 03:25, 23 January 2020
Line 1:		Line 1:
	Sumber: https://docs.biolab.si//3/visual-programming/widgets/model/stacking.html		Sumber: https://docs.biolab.si//3/visual-programming/widgets/model/stacking.html






Line 11:		Line 7:

	Data: input dataset		Data: input dataset

	Preprocessor: preprocessing method(s)		Preprocessor: preprocessing method(s)

	Learners: learning algorithm		Learners: learning algorithm

	Aggregate: model aggregation method		Aggregate: model aggregation method

Line 21:		Line 14:

	Learner: aggregated (stacked) learning algorithm		Learner: aggregated (stacked) learning algorithm

	Model: trained model		Model: trained model

	Stacking is an ensemble method that computes a meta model from several base models. The Stacking widget has the Aggregate input, which provides a method for aggregating the input models. If no aggregation input is given the default methods are used. Those are Logistic Regression for classification and Ridge Regression for regression problems.		Stacking is an ensemble method that computes a meta model from several base models. The Stacking widget has the Aggregate input, which provides a method for aggregating the input models. If no aggregation input is given the default methods are used. Those are Logistic Regression for classification and Ridge Regression for regression problems.

	~~../../_images/~~Stacking-stamped.png		[[File:Stacking-stamped.png\|center\|200px\|thumb]]

	The meta learner can be given a name under which it will appear in other widgets. The default name is “Stack”.		The meta learner can be given a name under which it will appear in other widgets. The default name is “Stack”.

	Click Apply to commit the aggregated model. 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.		Click Apply to commit the aggregated model. 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.

	Access help and produce a report.		Access help and produce a report.

	~~Example~~		==Contoh==

	We will use Paint Data to demonstrate how the widget is used. We painted a complex dataset with 4 class labels and sent it to Test & Score. We also provided three kNN learners, each with a different parameters (number of neighbors is 5, 10 or 15). Evaluation results are good, but can we do better?		We will use Paint Data to demonstrate how the widget is used. We painted a complex dataset with 4 class labels and sent it to Test & Score. We also provided three kNN learners, each with a different parameters (number of neighbors is 5, 10 or 15). Evaluation results are good, but can we do better?
Line 40:		Line 30:
	Let’s use Stacking. Stacking requires several learners on the input and an aggregation method. In our case, this is Logistic Regression. A constructed meta learner is then sent to Test & Score. Results have improved, even if only marginally. Stacking normally works well on complex data sets.		Let’s use Stacking. Stacking requires several learners on the input and an aggregation method. In our case, this is Logistic Regression. A constructed meta learner is then sent to Test & Score. Results have improved, even if only marginally. Stacking normally works well on complex data sets.

	~~../../_images/~~Stacking-Example.png		[[File:Stacking-Example.png\|center\|200px\|thumb]]

Onnowpurbo: Created page with "Sumber: https://docs.biolab.si//3/visual-programming/widgets/model/stacking.html Stack multiple models. Inputs Data: input dataset Preprocessor: preprocessing..."

2020-01-16T07:17:02Z

Created page with "Sumber: https://docs.biolab.si//3/visual-programming/widgets/model/stacking.html Stack multiple models. Inputs Data: input dataset Preprocessor: preprocessing..."

New page

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

Stack multiple models.

Inputs

Data: input dataset

Preprocessor: preprocessing method(s)

Learners: learning algorithm

Aggregate: model aggregation method

Outputs

Learner: aggregated (stacked) learning algorithm

Model: trained model

Stacking is an ensemble method that computes a meta model from several base models. The Stacking widget has the Aggregate input, which provides a method for aggregating the input models. If no aggregation input is given the default methods are used. Those are Logistic Regression for classification and Ridge Regression for regression problems.

../../_images/Stacking-stamped.png

The meta learner can be given a name under which it will appear in other widgets. The default name is “Stack”.

Click Apply to commit the aggregated model. 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.

Access help and produce a report.

Example

We will use Paint Data to demonstrate how the widget is used. We painted a complex dataset with 4 class labels and sent it to Test & Score. We also provided three kNN learners, each with a different parameters (number of neighbors is 5, 10 or 15). Evaluation results are good, but can we do better?

Let’s use Stacking. Stacking requires several learners on the input and an aggregation method. In our case, this is Logistic Regression. A constructed meta learner is then sent to Test & Score. Results have improved, even if only marginally. Stacking normally works well on complex data sets.

../../_images/Stacking-Example.png

==Referensi==

* https://docs.biolab.si//3/visual-programming/widgets/model/stacking.html

==Pranala Menarik==

* [[Orange]]