Vandelay Industries ! Probe File Exchange

View probe files

Select the checkbox beside one or more files to find the combined RMSE or to download the files.
Select a column heading to sort the entries.

    Check allUncheck all

File ID	User	Probe RMSE	Algorithm	Description
15	adifferent	0.870968	combination	Vandelay Industries ! Nonlinear ...
16	adifferent	0.878586	combination	OfADifferentKind mixture. Qualif...
17	chef-ele	0.88842	combination	A nonlinear blend of RBM, SVD, KN...
18	Newman	0.967224	combination	03.005_004
19	Newman	0.90724	combination	008_008
20	Newman	0.897456	combination	008_064
29	Newman	0.895524	combination	kl_016_080
31	Newman	0.98068	combination	lm2gedate
32	Newman	0.909343	combination	lm2rbm21
34	Newman	0.89921	combination	p__0004_0512
37	Newman	0.904472	combination	rbm24amf
39	Aron	0.873372	combination	50 predictor blend
43	soupnazi	0.891079	combination
59	DandA	0.887654	combination	My blend of SVD variants, RBM var...
66	PhanTom	0.870965	combination	another result based on "adiffere...
71	PhanTom	0.870968	combination	mainly based on "adifferent 0.870...
30	Aron	0.881453	hybrid	Single (not Bellkor) integrated m...
38	Aron	0.881733	hybrid	A logistic version of my integrat...
54	Aron	0.880893	hybrid	256 feature integrated model
27	Newman	0.94781	knn	d1.005_004
58	greymatter	0.883621	nsvd1	Single model result
40	Newman	0.87799	other	probe blend for my .8689 submissi...
42	soupnazi	0.91732	other
49	soupnazi	0.969917	other
52	soupnazi	0.951338	other
76	DandA	0.904528	other	Feedforward neural net with patte...
82	DandA	0.905643	other	Feedforward neural net, condition...
88	quadcore	0.901617	other	svd+++ with some neighbor thrown ...
22	Newman	0.9099	rbm	21
23	Newman	0.91297	rbm	22
35	Newman	0.899078	rbm	rbm21
36	Newman	0.898148	rbm	rbm21c
44	Aron	0.893186	rbm	This is an RBM with some date eff...
74	DandA	0.901696	rbm	Conditional factored RBM 1000-100
75	DandA	0.902159	rbm	CF RBM 1000-100
24	Newman	0.914631	svd	328-g2
25	Newman	0.922905	svd	328-g4
41	soupnazi	0.920143	svd
45	soupnazi	0.92168	svd
46	soupnazi	0.909504	svd
47	soupnazi	0.974801	svd
48	soupnazi	0.929019	svd
50	soupnazi	0.904926	svd
51	soupnazi	0.907313	svd
53	soupnazi	0.91286	svd
77	DandA	0.907028	svd	svd 500. Largest positive coeffic...
79	DandA	0.908265	svd	svd 250 plus biases
80	DandA	0.91197	svd	svd 40. Large negative coefficien...
21	Newman	0.89106	svd++	11
26	Newman	0.899632	svd++	1000
28	Newman	0.898801	svd++	e80i80edf
33	Newman	0.900418	svd++	m600
73	DandA	0.90424	svd++	svd++ 100
78	DandA	0.925821	svd++	svd++ 10. Large negative coeffici...
81	DandA	0.940254	svd++	svd++ 10 no biases. Large negativ...
94	kimclark	0.887506	svd++	Fully Bayesian treatment of a tim...

Vandelay Industries ! Nonlinear blend. Qualifying RMSE: 0.8664

OfADifferentKind mixture. Qualifying RMSE 0.8737

A nonlinear blend of RBM, SVD, KNN & other algorithms. It is now being used as part of Vandelay Industries

03.005_004

008_008

008_064

11

21

22

328-g2

328-g4

1000

d1.005_004

e80i80edf

kl_016_080

Single (not Bellkor) integrated model, using most known effects. Base:128f SVD+.

lm2gedate

lm2rbm21

m600

p__0004_0512

rbm21

rbm21c

rbm24amf

A logistic version of my integrated model.

50 predictor blend

probe blend for my .8689 submission

There is currently no description of this file.

There is currently no description of this file.

There is currently no description of this file.

This is an RBM with some date effects and a neighborhood-style post processor.

There is currently no description of this file.

There is currently no description of this file.

There is currently no description of this file.

There is currently no description of this file.

There is currently no description of this file.

There is currently no description of this file.

There is currently no description of this file.

There is currently no description of this file.

There is currently no description of this file.

256 feature integrated model

Single model result

My blend of SVD variants, RBM variants, plus custom models.

another result based on "adifferent 0.870968"

svd++ 100

Conditional factored RBM 1000-100

mainly based on "adifferent 0.870968" again

CF RBM 1000-100

Feedforward neural net with pattern-completion training. 400 hiddens

svd 500. Largest positive coefficient in my blend, but NOT the best individual model!

svd++ 10. Large negative coefficient in my blend. Very useful!

svd 250 plus biases

svd 40. Large negative coefficient in my blend. very useful

svd++ 10 no biases. Large negative coefficient in my blend. very useful.

Feedforward neural net, conditional, 100 hiddens

svd+++ with some neighbor thrown in. poor RMSE but blends well

Fully Bayesian treatment of a time integrated svd++ model