Blending layers of CUMULATE's user model - Revision history

Myudelson: /* Publication */

2009-06-28T21:04:51Z

‎Publication

Myudelson: /* Publication */

2009-06-28T21:04:38Z

‎Publication

Myudelson at 21:00, 28 June 2009

2009-06-28T21:00:51Z

Myudelson: /* Publication */

2009-06-28T21:00:00Z

‎Publication

Myudelson at 20:59, 28 June 2009

2009-06-28T20:59:46Z

Myudelson: /* Results */

2009-06-28T20:14:17Z

‎Results

Myudelson at 20:13, 28 June 2009

2009-06-28T20:13:53Z

Myudelson at 20:10, 28 June 2009

2009-06-28T20:10:48Z

Myudelson: /* Results */

2009-06-28T14:30:37Z

‎Results

Myudelson: /* Results */

2009-06-28T14:30:00Z

‎Results

@@ Line 82: / Line 82: @@
 === Publication ===
-Yudelson, M., Sosnovsky, A. (2009). Retrospective evaluation of blended user modeling for adaptive educational systems. In: Weibelzahl, S., Masthoff., J., Paramythis, A., van Velsen, L. (Eds.) [http://www.easy-hub.org/hub/workshops/umap2009/index.html 6th Workshop on User-Centred Design and Evaluation of Adaptive Systems (UCDEAS)], pp. 34-43. June 26, 2009. Trento, Italy. [[Media:UMAP09.UCDEAS.Yudelson.Retrosp. Eval. Blended UM For AES.pdf|slides]]
+Yudelson, M., Sosnovsky, A. (2009). Retrospective evaluation of blended user modeling for adaptive educational systems. In: Weibelzahl, S., Masthoff., J., Paramythis, A., van Velsen, L. (Eds.) [http://www.easy-hub.org/hub/workshops/umap2009/index.html 6th Workshop on User-Centred Design and Evaluation of Adaptive Systems (UCDEAS)], pp. 34-43. June 26, 2009. Trento, Italy. ([[Media:UMAP09.UCDEAS.Yudelson.Retrosp. Eval. Blended UM For AES.pdf|slides]])
 === Addendum ===

@@ Line 82: / Line 82: @@
 === Publication ===
-Yudelson, M., Sosnovsky, A. (2009). Retrospective evaluation of blended user modeling for adaptive educational systems. In: Weibelzahl, S., Masthoff., J., Paramythis, A., van Velsen, L. (Eds.) 6rd Workshop on User-Centred Design and Evaluation of Adaptive Systems (UCDEAS), pp. 34-43. June 26, 2009. Trento, Italy. [[Media:UMAP09.UCDEAS.Yudelson.Retrosp. Eval. Blended UM For AES.pdf|slides]]
+Yudelson, M., Sosnovsky, A. (2009). Retrospective evaluation of blended user modeling for adaptive educational systems. In: Weibelzahl, S., Masthoff., J., Paramythis, A., van Velsen, L. (Eds.) [http://www.easy-hub.org/hub/workshops/umap2009/index.html 6th Workshop on User-Centred Design and Evaluation of Adaptive Systems (UCDEAS)], pp. 34-43. June 26, 2009. Trento, Italy. [[Media:UMAP09.UCDEAS.Yudelson.Retrosp. Eval. Blended UM For AES.pdf|slides]]
 === Addendum ===

← Older revision		Revision as of 21:00, 28 June 2009
Line 88:		Line 88:

	[[CUMULATE]] log data for the 6 Database Management Courses can be accessed [[CUMULATE_Usage_Logs#Database_Management_6_Course_Dataset\|here]]		[[CUMULATE]] log data for the 6 Database Management Courses can be accessed [[CUMULATE_Usage_Logs#Database_Management_6_Course_Dataset\|here]]
		+
		+	= Contacts =
		+	[[User:Myudelson\|Michael V. Yudelson]]

@@ Line 82: / Line 82: @@
 === Publication ===
-Yudelson, M., Sosnovsky, A. (2009). Retrospective evaluation of blended user modeling for adaptive educational systems. In: Weibelzahl, S., Masthoff., J., Paramythis, A., van Velsen, L. (Eds.) 6rd Workshop on User-Centred Design and Evaluation of Adaptive Systems (UCDEAS), pp. 34-43. June 26, 2009. Trento, Italy. [[Media:UMAP09.UCDEAS.Yudelson.Retrosp. Eval. Blended UM For AES.pdfslides]]
+Yudelson, M., Sosnovsky, A. (2009). Retrospective evaluation of blended user modeling for adaptive educational systems. In: Weibelzahl, S., Masthoff., J., Paramythis, A., van Velsen, L. (Eds.) 6rd Workshop on User-Centred Design and Evaluation of Adaptive Systems (UCDEAS), pp. 34-43. June 26, 2009. Trento, Italy. [[Media:UMAP09.UCDEAS.Yudelson.Retrosp. Eval. Blended UM For AES.pdf|slides]]
 === Addendum ===

← Older revision		Revision as of 20:59, 28 June 2009
Line 80:		Line 80:
	** '''''High'''''. 5-9 best blends. 17 users. This group follows the global trend with 40% and 50% blends having best significant edge over 0% blend: p-value = .049		** '''''High'''''. 5-9 best blends. 17 users. This group follows the global trend with 40% and 50% blends having best significant edge over 0% blend: p-value = .049
	** <span style="color:grey;">'''''No difference'''''. 10 best blends (all blended models have same accuracy). 15 users. No further investigation performed.</span>		** <span style="color:grey;">'''''No difference'''''. 10 best blends (all blended models have same accuracy). 15 users. No further investigation performed.</span>
		+
		+	=== Publication ===
		+	Yudelson, M., Sosnovsky, A. (2009). Retrospective evaluation of blended user modeling for adaptive educational systems. In: Weibelzahl, S., Masthoff., J., Paramythis, A., van Velsen, L. (Eds.) 6rd Workshop on User-Centred Design and Evaluation of Adaptive Systems (UCDEAS), pp. 34-43. June 26, 2009. Trento, Italy. [[Media:UMAP09.UCDEAS.Yudelson.Retrosp. Eval. Blended UM For AES.pdfslides]]
		+
		+	=== Addendum ===
		+	MATLAB code used for computation can be downloaded [[Media:CUMULATE blending user model layers study1 umap09 ucdeas matlab.src.tgz\|here]].
		+
		+	[[CUMULATE]] log data for the 6 Database Management Courses can be accessed [[CUMULATE_Usage_Logs#Database_Management_6_Course_Dataset\|here]]

@@ Line 71: / Line 71: @@
 === Results ===
-[[Image:Blending layers of CUMULATE's user model. Study 1.users.png|thumb|'''100'''|Example of blend accuracies for 5 users]]
+[[Image:Blending layers of CUMULATE's user model. Study 1.users.png|thumb|'''100'''|Examples of blend effect on user model accuracy]]
 [[Image:Blending layers of CUMULATE's user model. Study 1.histogram.png|thumb|'''100'''|Distribution of number of users for different "best" blends]]
 * While looking at users' individual preferences, the best user blend seems to be superior to the 0%-blend: t = -5.38, p-value<.001

@@ Line 71: / Line 71: @@
 === Results ===
 * While looking at users' individual preferences, the best user blend seems to be superior to the 0%-blend: t = -5.38, p-value<.001
 * Comparison of uniform blends for all users to a 0%-blends, shows that 40% and 50% blends were significantly better than 0%-blend: t = -for both 2.05, p-value = .023
@@ Line 78: / Line 80: @@
 ** '''''High'''''. 5-9 best blends. 17 users. This group follows the global trend with 40% and 50% blends having best significant edge over 0% blend: p-value = .049
 ** <span style="color:grey;">'''''No difference'''''. 10 best blends (all blended models have same accuracy). 15 users. No further investigation performed.</span>

← Older revision		Revision as of 20:10, 28 June 2009
Line 78:		Line 78:
	** '''''High'''''. 5-9 best blends. 17 users. This group follows the global trend with 40% and 50% blends having best significant edge over 0% blend: p-value = .049		** '''''High'''''. 5-9 best blends. 17 users. This group follows the global trend with 40% and 50% blends having best significant edge over 0% blend: p-value = .049
	** <span style="color:grey;">'''''No difference'''''. 10 best blends (all blended models have same accuracy). 15 users. No further investigation performed.</span>		** <span style="color:grey;">'''''No difference'''''. 10 best blends (all blended models have same accuracy). 15 users. No further investigation performed.</span>
		+
		+	[[Image:Blending layers of CUMULATE's user model. Study 1.histogram.png]]
		+	[[Image:Blending layers of CUMULATE's user model. Study 1.users.png]]

@@ Line 74: / Line 74: @@
 * Comparison of uniform blends for all users to a 0%-blends, shows that 40% and 50% blends were significantly better than 0%-blend: t = -for both 2.05, p-value = .023
 * User separation into groups by the number of best blends yielded the following results
-** <span style="color:grey;">"Low". 1 best blend only, 2 users. No further investigation performed.</span>
+** <span style="color:grey;">'''''Low'''''. 1 best blend only, 2 users. No further investigation performed.</span>
-** "Medium". 2-4 best blends. 22 users. Here 90% blend had the highest significant edge over 0% blend: p-value = .037
+** '''''Medium'''''. 2-4 best blends. 22 users. Here 90% blend had the highest significant edge over 0% blend: p-value = .037
-** "High". 5-9 best blends. 17 users. This group follows the global trend with 40% and 50% blends having best significant edge over 0% blend: p-value = .049
+** '''''High'''''. 5-9 best blends. 17 users. This group follows the global trend with 40% and 50% blends having best significant edge over 0% blend: p-value = .049
-** <span style="color:grey;">"No difference". 10 best blends (all blended models have same accuracy). 15 users. No further investigation performed.</span>
+** <span style="color:grey;">'''''No difference'''''. 10 best blends (all blended models have same accuracy). 15 users. No further investigation performed.</span>