1 FRESA.CAD Regresion Benchmark

1.1 BRCA Recurence RISK Data Set


lesionsSumDiffRed <- NULL
load("RadiomicsBRCA.RDATA")



ExperimentName <- "HSJBC_Oncotype"
bswimsReps <- 20;
theData <- lesionsSumDiffRed;
theOutcome <- "Risk";
reps <- 100;
fraction <- 0.9;

FRESAFileName <- paste(ExperimentName,"FRESAMethod.RDATA",sep = "_")
CVFileName <- paste(ExperimentName,"CVMethod.RDATA",sep = "_")

1.2 Benchmark


MODELFRESA <- FRESA.Model(formula = paste(theOutcome," ~ 1"),data = theData,repeats = bswimsReps)
save(MODELFRESA,file= FRESAFileName)

cp <- RegresionBenchmark(theData = theData, theOutcome = theOutcome, reps = reps, fraction = fraction)
save(cp,file = CVFileName)



elapcol <- names(cp$times[[1]]) == "elapsed"
cputimes <- list(Fresa = mean(cp$times$BSWiMS[ elapcol ]),LASSO = mean(cp$times$LASSO[ elapcol ]),RF = mean(cp$times$RF[ elapcol ]),SVM = mean(cp$times$SVM[ elapcol ]))

featsize <- list(BSWiMS = cp$jaccard_filter$BSWiMS$averageLength,
                 LASSO = cp$jaccard_filter$LASSO$averageLength,
                 RPART= cp$jaccard_filter$rpart$averageLength,
                 RF = cp$jaccard_filter$RF$averageLength,
                 FT = cp$jaccard_filter$FT$averageLength,
                 TT = cp$jaccard_filter$TT$averageLength,
                 WT = cp$jaccard_filter$WT$averageLength,
                 BF = cp$jaccard_filter$BT$averageLength,
                 pearson = cp$jaccard_filter$pearson$averageLength,
                 spearman = cp$jaccard_filter$spearman$averageLength,
                 kendall = cp$jaccard_filter$kendall$averageLength,
                 mRMR = cp$jaccard_filter$mRMR$averageLength
                 )

jaccard <- list(BSWiMS = cp$jaccard_filter$BSWiMS$Jaccard.SM,
                 LASSO = cp$jaccard_filter$LASSO$Jaccard.SM,
                 RPART = cp$jaccard_filter$rpart$Jaccard.SM,
                 RF = cp$jaccard_filter$RF$Jaccard.SM,
                 FT = cp$jaccard_filter$FT$Jaccard.SM,
                 TT = cp$jaccard_filter$TT$Jaccard.SM,
                 WT = cp$jaccard_filter$WT$Jaccard.SM,
                 BT = cp$jaccard_filter$BT$Jaccard.SM,
                 pearson = cp$jaccard_filter$pearson$Jaccard.SM,
                 spearman = cp$jaccard_filter$spearman$Jaccard.SM,
                 kendall = cp$jaccard_filter$kendall$Jaccard.SM,
                 mRMR = cp$jaccard_filter$mRMR$Jaccard.SM
                 )

1.2.1 Results

#The Times
pander::pander(cputimes)

Fresa: 0.6421
LASSO: 0.2318
RF: 1.762
SVM: 0.0098

pander::pander(featsize)

BSWiMS: 6.53
LASSO: 16.84
RPART: 24.95
RF: 6.65
FT: 8.15
TT: 0
WT: 0
BF: 0
pearson: 2.7
spearman: 4.8
kendall: 4.02
mRMR: 31




bp <- barPlotCiError(as.matrix(cp$CorTable),metricname = "Pearson Correlation",thesets = cp$thesets,themethod = cp$theMethod,main = "Pearson Correlation",offsets = c(0.5,0.05),args.legend = list(x = "bottomright"))


pander::pander(bp$barMatrix,caption = "Pearson Correlation",round = 3)

Pearson Correlation
	Regression Algorithm
BSWiMS	0.403
RF	0.218
RPART	0.113
LASSO	0.398
SVM	0.381
ENS	0.398

pander::pander(bp$ciTable,caption = "Pearson Correlation with 95%CI",round = 3)

Pearson Correlation with 95%CI
	Pearson Correlation	lower	upper
Regression Algorithm	0.403	0.187	0.582
Regression Algorithm	0.218	-0.017	0.429
Regression Algorithm	0.113	-0.124	0.337
Regression Algorithm	0.398	0.182	0.578
Regression Algorithm	0.381	0.162	0.564
Regression Algorithm	0.398	0.182	0.578



bp <- barPlotCiError(as.matrix(cp$RMSETable),metricname = "RMSE",thesets = cp$thesets,themethod = cp$theMethod,main = "RMSE",offsets = c(0.5,5),args.legend = list(x = "bottomright"))

pander::pander(bp$barMatrix,caption = "RMSE",round = 3)

RMSE
	Regression Algorithm
BSWiMS	50.71
RF	54.07
RPART	61.35
LASSO	50.75
SVM	51.28
ENS	50.76

pander::pander(bp$ciTable,caption = "RMSE with 95%CI",round = 3)

RMSE with 95%CI
	RMSE	lower	upper
Regression Algorithm	50.71	43.57	60.68
Regression Algorithm	54.07	46.46	64.7
Regression Algorithm	61.35	52.71	73.41
Regression Algorithm	50.75	43.6	60.72
Regression Algorithm	51.28	44.06	61.36
Regression Algorithm	50.76	43.61	60.74


  
bp <- barPlotCiError(as.matrix(cp$BiasTable),metricname = "BIAS",thesets = cp$thesets,themethod = cp$theMethod,main = "BIAS",offsets = c(0.5,0.5),args.legend = list(x = "bottomright"))

pander::pander(bp$barMatrix,caption = "BIAS",round = 3)

BIAS
	Regression Algorithm
BSWiMS	-0.117
RF	0.615
RPART	-2.126
LASSO	0.636
SVM	-3.262
ENS	-0.532

pander::pander(bp$ciTable,caption = "BIAS with 95%CI",round = 3)

BIAS with 95%CI
	BIAS	lower	upper
Regression Algorithm	-0.117	-12.12	11.89
Regression Algorithm	0.615	-12.18	13.41
Regression Algorithm	-2.126	-16.64	12.39
Regression Algorithm	0.636	-11.38	12.65
Regression Algorithm	-3.262	-15.38	8.851
Regression Algorithm	-0.532	-12.55	11.48



barplot(unlist(jaccard),las = 2,cex.axis = 1,cex.names = 0.7,main = "Jaccard Index",xlab = "Filter Method",ylab = "Jaccard")

barplot(unlist(featsize)+1,las = 2,cex.axis = 1,cex.names = 0.7,log = "y",xlab = "Filter Method",ylab = "# of Features+1")




bp <- barPlotCiError(as.matrix(cp$CorTable_filter),metricname = "Pearson Correlation",thesets = cp$theFiltersets,themethod = cp$theRegressMethod,main = "Pearson Correlation",offsets = c(0.0,0.10),args.legend = list(x = "bottomright"),angle=45)


pander::pander(bp$barMatrix,caption = "Pearson Correlation",round = 3)

Pearson Correlation (continued below)
	BSWiMS	LASSO	RPART	RF:BSWiMS	F-Test	t-Test
Linear Regression	0.432	0.459	0.385	0.356	0.465	-0.685
Robust Regression	0.422	0.445	0.371	0.389	0.454	-0.685
Ridge Regression	0.434	0.462	0.412	0.355	0.463	-0.685
LASSO	0.431	0.433	0.418	0.336	0.424	-0.685
SVM	0.471	0.41	0.323	0.453	0.382	-0.685
Random Forest	0.422	0.381	0.26	0.339	0.347	-0.685

Table continues below
	W-Test	Binomial Test	Pearson	Spearman	Kendall
Linear Regression	-0.685	-0.685	0.345	0.369	0.368
Robust Regression	-0.685	-0.685	0.336	0.381	0.383
Ridge Regression	-0.685	-0.685	0.344	0.376	0.381
LASSO	-0.685	-0.685	0.346	0.362	0.366
SVM	-0.685	-0.685	0.446	0.382	0.354
Random Forest	-0.685	-0.685	0.338	0.332	0.32

	mRMR
Linear Regression	0.417
Robust Regression	0.385
Ridge Regression	0.414
LASSO	0.401
SVM	0.381
Random Forest	0.337

pander::pander(bp$ciTable,caption = "Pearson Correlation with 95%CI",round = 3)

Pearson Correlation with 95%CI
	Pearson Correlation	lower	upper
BSWiMS	0.432	0.221	0.605
BSWiMS	0.422	0.209	0.597
BSWiMS	0.434	0.224	0.606
BSWiMS	0.431	0.22	0.604
BSWiMS	0.471	0.267	0.634
BSWiMS	0.422	0.209	0.596
LASSO	0.459	0.253	0.626
LASSO	0.445	0.236	0.614
LASSO	0.462	0.256	0.628
LASSO	0.433	0.222	0.605
LASSO	0.41	0.195	0.587
LASSO	0.381	0.162	0.564
RPART	0.385	0.167	0.567
RPART	0.371	0.151	0.556
RPART	0.412	0.198	0.589
RPART	0.418	0.204	0.593
RPART	0.323	0.097	0.517
RPART	0.26	0.028	0.465
RF:BSWiMS	0.356	0.134	0.544
RF:BSWiMS	0.389	0.171	0.57
RF:BSWiMS	0.355	0.133	0.543
RF:BSWiMS	0.336	0.112	0.528
RF:BSWiMS	0.453	0.245	0.62
RF:BSWiMS	0.339	0.115	0.53
F-Test	0.465	0.259	0.63
F-Test	0.454	0.246	0.621
F-Test	0.463	0.258	0.629
F-Test	0.424	0.212	0.598
F-Test	0.382	0.163	0.565
F-Test	0.347	0.124	0.537
t-Test	-0.685	-0.792	-0.538
t-Test	-0.685	-0.792	-0.538
t-Test	-0.685	-0.792	-0.538
t-Test	-0.685	-0.792	-0.538
t-Test	-0.685	-0.792	-0.538
t-Test	-0.685	-0.792	-0.538
W-Test	-0.685	-0.792	-0.538
W-Test	-0.685	-0.792	-0.538
W-Test	-0.685	-0.792	-0.538
W-Test	-0.685	-0.792	-0.538
W-Test	-0.685	-0.792	-0.538
W-Test	-0.685	-0.792	-0.538
Binomial Test	-0.685	-0.792	-0.538
Binomial Test	-0.685	-0.792	-0.538
Binomial Test	-0.685	-0.792	-0.538
Binomial Test	-0.685	-0.792	-0.538
Binomial Test	-0.685	-0.792	-0.538
Binomial Test	-0.685	-0.792	-0.538
Pearson	0.345	0.121	0.535
Pearson	0.336	0.112	0.528
Pearson	0.344	0.12	0.534
Pearson	0.346	0.123	0.536
Pearson	0.446	0.238	0.616
Pearson	0.338	0.113	0.529
Spearman	0.369	0.149	0.555
Spearman	0.381	0.162	0.564
Spearman	0.376	0.157	0.56
Spearman	0.362	0.14	0.549
Spearman	0.382	0.163	0.565
Spearman	0.332	0.107	0.525
Kendall	0.368	0.147	0.554
Kendall	0.383	0.165	0.566
Kendall	0.381	0.162	0.564
Kendall	0.366	0.146	0.553
Kendall	0.354	0.131	0.542
Kendall	0.32	0.094	0.515
mRMR	0.417	0.204	0.593
mRMR	0.385	0.167	0.567
mRMR	0.414	0.2	0.59
mRMR	0.401	0.185	0.58
mRMR	0.381	0.162	0.564
mRMR	0.337	0.113	0.529



bp <- barPlotCiError(as.matrix(cp$RMSETable_filter),metricname = "RMSE",thesets = cp$theFiltersets,themethod = cp$theRegressMethod,main = "RMSE",offsets = c(0.0,10.0),args.legend = list(x = "bottomright"),angle=45)

pander::pander(bp$barMatrix,caption = "RMSE",round = 3)

RMSE (continued below)
	BSWiMS	LASSO	RPART	RF:BSWiMS	F-Test	t-Test
Linear Regression	51.03	51.09	53.97	53.54	49.68	56.33
Robust Regression	51.85	51.7	55.2	52.58	50.31	56.33
Ridge Regression	50.84	50.95	52.22	53.59	49.67	56.33
LASSO	50.5	52.3	50.33	53.07	50.55	56.33
SVM	49.66	50.66	52.73	50.02	52.48	56.33
Random Forest	50.42	51.14	54.11	52.78	52.29	56.33

Table continues below
	W-Test	Binomial Test	Pearson	Spearman	Kendall
Linear Regression	56.33	56.33	52.85	52.38	52.37
Robust Regression	56.33	56.33	53.45	52.3	52.3
Ridge Regression	56.33	56.33	52.88	52.1	51.77
LASSO	56.33	56.33	52.69	52.48	52.09
SVM	56.33	56.33	50.14	52.16	52.91
Random Forest	56.33	56.33	52.52	52.44	52.92

	mRMR
Linear Regression	58.39
Robust Regression	62.11
Ridge Regression	58.25
LASSO	53.47
SVM	51.28
Random Forest	52.2

pander::pander(bp$ciTable,caption = "RMSE with 95%CI",round = 3)

RMSE with 95%CI
	RMSE	lower	upper
BSWiMS	51.03	43.85	61.06
BSWiMS	51.85	44.55	62.05
BSWiMS	50.84	43.68	60.84
BSWiMS	50.5	43.39	60.43
BSWiMS	49.66	42.67	59.42
BSWiMS	50.42	43.31	60.33
LASSO	51.09	43.89	61.13
LASSO	51.7	44.42	61.87
LASSO	50.95	43.77	60.96
LASSO	52.3	44.93	62.57
LASSO	50.66	43.53	60.62
LASSO	51.14	43.93	61.19
RPART	53.97	46.37	64.57
RPART	55.2	47.43	66.05
RPART	52.22	44.87	62.48
RPART	50.33	43.24	60.22
RPART	52.73	45.3	63.1
RPART	54.11	46.49	64.74
RF:BSWiMS	53.54	46	64.07
RF:BSWiMS	52.58	45.18	62.92
RF:BSWiMS	53.59	46.04	64.12
RF:BSWiMS	53.07	45.6	63.5
RF:BSWiMS	50.02	42.97	59.85
RF:BSWiMS	52.78	45.34	63.15
F-Test	49.68	42.68	59.44
F-Test	50.31	43.23	60.2
F-Test	49.67	42.67	59.43
F-Test	50.55	43.43	60.48
F-Test	52.48	45.09	62.79
F-Test	52.29	44.92	62.56
t-Test	56.33	48.4	67.4
t-Test	56.33	48.4	67.4
t-Test	56.33	48.4	67.4
t-Test	56.33	48.4	67.4
t-Test	56.33	48.4	67.4
t-Test	56.33	48.4	67.4
W-Test	56.33	48.4	67.4
W-Test	56.33	48.4	67.4
W-Test	56.33	48.4	67.4
W-Test	56.33	48.4	67.4
W-Test	56.33	48.4	67.4
W-Test	56.33	48.4	67.4
Binomial Test	56.33	48.4	67.4
Binomial Test	56.33	48.4	67.4
Binomial Test	56.33	48.4	67.4
Binomial Test	56.33	48.4	67.4
Binomial Test	56.33	48.4	67.4
Binomial Test	56.33	48.4	67.4
Pearson	52.85	45.4	63.23
Pearson	53.45	45.92	63.95
Pearson	52.88	45.43	63.27
Pearson	52.69	45.27	63.05
Pearson	50.14	43.08	60
Pearson	52.52	45.12	62.85
Spearman	52.38	45.01	62.68
Spearman	52.3	44.93	62.58
Spearman	52.1	44.76	62.34
Spearman	52.48	45.09	62.8
Spearman	52.16	44.81	62.41
Spearman	52.44	45.05	62.74
Kendall	52.37	45	62.66
Kendall	52.3	44.94	62.58
Kendall	51.77	44.48	61.94
Kendall	52.09	44.75	62.32
Kendall	52.91	45.45	63.3
Kendall	52.92	45.47	63.32
mRMR	58.39	50.17	69.86
mRMR	62.11	53.36	74.31
mRMR	58.25	50.04	69.7
mRMR	53.47	45.94	63.97
mRMR	51.28	44.05	61.36
mRMR	52.2	44.85	62.46


  

bp <- barPlotCiError(as.matrix(cp$BiasTable_filter),metricname = "BIAS",thesets = cp$theFiltersets,themethod = cp$theRegressMethod,main = "BIAS",offsets = c(0.5,5),args.legend = list(x = "topleft"),angle=45)

pander::pander(bp$barMatrix,caption = "BIAS",round = 3)

BIAS (continued below)
	BSWiMS	LASSO	RPART	RF:BSWiMS	F-Test	t-Test
Linear Regression	-0.209	0.316	3.051	-0.94	0.16	-0.058
Robust Regression	0.921	1.2	3.51	-1.454	1.613	-0.058
Ridge Regression	0.019	0.154	3.016	-0.541	0.012	-0.058
LASSO	0.079	0.621	1.878	-0.824	-0.928	-0.058
SVM	-0.671	-1.691	1.035	0.479	-2.427	-0.058
Random Forest	-1.52	-0.5	0.359	0.182	-1.843	-0.058

Table continues below
	W-Test	Binomial Test	Pearson	Spearman	Kendall
Linear Regression	-0.058	-0.058	0.047	-0.957	0.627
Robust Regression	-0.058	-0.058	1.482	0.635	1.957
Ridge Regression	-0.058	-0.058	0.071	-0.965	0.097
LASSO	-0.058	-0.058	-0.087	-0.968	-0.296
SVM	-0.058	-0.058	2.567	0.967	0.82
Random Forest	-0.058	-0.058	-0.252	-0.721	-1.711

	mRMR
Linear Regression	3.348
Robust Regression	4.06
Ridge Regression	3.316
LASSO	0.647
SVM	-3.262
Random Forest	0.166

pander::pander(bp$ciTable,caption = "BIAS with 95%CI",round = 3)

BIAS with 95%CI
	BIAS	lower	upper
BSWiMS	-0.209	-12.29	11.87
BSWiMS	0.921	-11.35	13.19
BSWiMS	0.019	-12.02	12.05
BSWiMS	0.079	-11.88	12.03
BSWiMS	-0.671	-12.43	11.08
BSWiMS	-1.52	-13.45	10.41
LASSO	0.316	-11.78	12.41
LASSO	1.2	-11.04	13.44
LASSO	0.154	-11.91	12.21
LASSO	0.621	-11.76	13
LASSO	-1.691	-13.68	10.29
LASSO	-0.5	-12.6	11.6
RPART	3.051	-9.703	15.8
RPART	3.51	-9.53	16.55
RPART	3.016	-9.325	15.36
RPART	1.878	-10.03	13.78
RPART	1.035	-11.44	13.51
RPART	0.359	-12.45	13.16
RF:BSWiMS	-0.94	-13.61	11.73
RF:BSWiMS	-1.454	-13.9	10.99
RF:BSWiMS	-0.541	-13.22	12.14
RF:BSWiMS	-0.824	-13.38	11.74
RF:BSWiMS	0.479	-11.36	12.32
RF:BSWiMS	0.182	-12.31	12.67
F-Test	0.16	-11.6	11.92
F-Test	1.613	-10.29	13.52
F-Test	0.012	-11.74	11.77
F-Test	-0.928	-12.89	11.04
F-Test	-2.427	-14.84	9.981
F-Test	-1.843	-14.21	10.53
t-Test	-0.058	-13.39	13.28
t-Test	-0.058	-13.39	13.28
t-Test	-0.058	-13.39	13.28
t-Test	-0.058	-13.39	13.28
t-Test	-0.058	-13.39	13.28
t-Test	-0.058	-13.39	13.28
W-Test	-0.058	-13.39	13.28
W-Test	-0.058	-13.39	13.28
W-Test	-0.058	-13.39	13.28
W-Test	-0.058	-13.39	13.28
W-Test	-0.058	-13.39	13.28
W-Test	-0.058	-13.39	13.28
Binomial Test	-0.058	-13.39	13.28
Binomial Test	-0.058	-13.39	13.28
Binomial Test	-0.058	-13.39	13.28
Binomial Test	-0.058	-13.39	13.28
Binomial Test	-0.058	-13.39	13.28
Binomial Test	-0.058	-13.39	13.28
Pearson	0.047	-12.46	12.56
Pearson	1.482	-11.16	14.13
Pearson	0.071	-12.45	12.59
Pearson	-0.087	-12.56	12.38
Pearson	2.567	-9.287	14.42
Pearson	-0.252	-12.68	12.18
Spearman	-0.957	-13.35	11.44
Spearman	0.635	-11.74	13.01
Spearman	-0.965	-13.29	11.36
Spearman	-0.968	-13.39	11.45
Spearman	0.967	-11.38	13.31
Spearman	-0.721	-13.13	11.69
Kendall	0.627	-11.77	13.02
Kendall	1.957	-10.41	14.33
Kendall	0.097	-12.16	12.35
Kendall	-0.296	-12.62	12.03
Kendall	0.82	-11.7	13.34
Kendall	-1.711	-14.23	10.81
mRMR	3.348	-10.45	17.15
mRMR	4.06	-10.61	18.73
mRMR	3.316	-10.45	17.08
mRMR	0.647	-12.01	13.3
mRMR	-3.262	-15.38	8.851
mRMR	0.166	-12.19	12.52

1.3 Features Analysis

#load(file = CVFileName)

ff <- names(cp$TheCVEvaluations$BSWIMS$featureFrequency)
ff <- c(ff,names(cp$TheCVEvaluations$LASSO$featureFrequency))
ff <- c(ff,names(cp$TheCVEvaluations$RPART$featureFrequency))
ff <- c(ff,names(cp$TheCVEvaluations$FRF$featureFrequency))
ff <- c(ff,names(cp$TheCVEvaluations$FT$featureFrequency))
ff <- c(ff,names(cp$TheCVEvaluations$TT$featureFrequency))
ff <- c(ff,names(cp$TheCVEvaluations$WT$featureFrequency))
ff <- c(ff,names(cp$TheCVEvaluations$BT$featureFrequency))
ff <- c(ff,names(cp$TheCVEvaluations$pearson$featureFrequency))
ff <- c(ff,names(cp$TheCVEvaluations$kendall$featureFrequency))
ff <- c(ff,names(cp$TheCVEvaluations$spearman$featureFrequency))
ff <- c(ff,names(cp$TheCVEvaluations$mRMR$featureFrequency))
ff <- unique(ff)

Nvar <- min(c(1000,length(ff)))
selFrequency <- matrix(0,nrow=Nvar,ncol=length(cp$theFiltersets))
rownames(selFrequency) <- names(cp$TheCVEvaluations$RF$featureFrequency)[1:Nvar]
selnames <- rownames(selFrequency)
colnames(selFrequency) <- cp$theFiltersets
ff <- cp$TheCVEvaluations$BSWIMS$featureFrequency
fnames <- selnames %in% names(ff)
selFrequency[fnames,"BSWiMS"] <- ff[selnames[fnames]]
ff <- cp$TheCVEvaluations$LASSO$featureFrequency
fnames <- selnames %in% names(ff)
selFrequency[fnames,"LASSO"] <- ff[selnames[fnames]]
ff <- cp$TheCVEvaluations$RPART$featureFrequency
fnames <- selnames %in% names(ff)
selFrequency[fnames,"RPART"] <- ff[selnames[fnames]]
ff <- cp$TheCVEvaluations$FRF$featureFrequency
fnames <- selnames %in% names(ff)
selFrequency[fnames,"RF:BSWiMS"] <- ff[selnames[fnames]]
ff <- cp$TheCVEvaluations$FT$featureFrequency
fnames <- selnames %in% names(ff)
selFrequency[fnames,"F-Test"] <- ff[selnames[fnames]]
ff <- cp$TheCVEvaluations$TT$featureFrequency
fnames <- selnames %in% names(ff)
selFrequency[fnames,"t-Test"] <- ff[selnames[fnames]]
ff <- cp$TheCVEvaluations$WT$featureFrequency
fnames <- selnames %in% names(ff)
selFrequency[fnames,"W-Test"] <- ff[selnames[fnames]]
ff <- cp$TheCVEvaluations$BT$featureFrequency
fnames <- selnames %in% names(ff)
selFrequency[fnames,"Binomial Test"] <- ff[selnames[fnames]]
ff <- cp$TheCVEvaluations$pearson$featureFrequency
fnames <- selnames %in% names(ff)
selFrequency[fnames,"Pearson"] <- ff[selnames[fnames]]
ff <- cp$TheCVEvaluations$kendall$featureFrequency
fnames <- selnames %in% names(ff)
selFrequency[fnames,"Kendall"] <- ff[selnames[fnames]]
ff <- cp$TheCVEvaluations$spearman$featureFrequency
fnames <- selnames %in% names(ff)
selFrequency[fnames,"Spearman"] <- ff[selnames[fnames]]
ff <- cp$TheCVEvaluations$mRMR$featureFrequency
fnames <- selnames %in% names(ff)
selFrequency[fnames,"mRMR"] <- ff[selnames[fnames]]
selFrequency <- selFrequency/reps
rm <- rowMeans(selFrequency)
selFrequency <- selFrequency[rm>0.02,]
gplots::heatmap.2(selFrequency,trace="none",mar=c(10,10),main="Features",cexRow = 0.6)


pander::pander(summary(MODELFRESA$BSWiMS.model,caption="Model",round = 3))

coefficients:

Table continues below
	Estimate	lower	mean	upper
D_CC_LH_3_ICDF_0.99	-0.5748	-0.7162	-0.5748	-0.4334
D_CC_HL_4_ICDF_0.999	-0.06073	-0.06823	-0.06073	-0.05323
S_CC_GLCM_4_Inertia	-5.84	-8.24	-5.84	-3.44
S_CC_FRACTAL_ICDF_0.95	0.1345	0.1272	0.1345	0.1418
D_CC_LH_3_ICDF_0.999	-0.04685	-0.04773	-0.04685	-0.04597
S_CC_FRACTAL_ICDF_0.75	0.007493	0.003408	0.007493	0.01158
S_CC_LH_2_Mean	-0.172	-0.2941	-0.172	-0.04986
S_CC_FRACTAL_Mean	0.01165	0.005211	0.01165	0.01809
S_CC_FRACTAL_ICDF_0.25	0.001734	0.0006536	0.001734	0.002815

Table continues below
	u.MSE	r.MSE	model.MSE	NeRI	F.pvalue
D_CC_LH_3_ICDF_0.99	2443	2140	1594	0.2113	2.677e-06
D_CC_HL_4_ICDF_0.999	2750	1998	1594	0.07042	3.226e-05
S_CC_GLCM_4_Inertia	2684	2144	1553	0.4085	9.27e-07
S_CC_FRACTAL_ICDF_0.95	2422	2304	1820	0.3114	1.997e-05
D_CC_LH_3_ICDF_0.999	2503	2456	2090	0.145	0.0003702
S_CC_FRACTAL_ICDF_0.75	2483	2844	2360	0.167	0.0001565
S_CC_LH_2_Mean	2586	2508	2251	0.1268	0.002861
S_CC_FRACTAL_Mean	2473	2627	2193	0.1956	0.0001926
S_CC_FRACTAL_ICDF_0.25	2535	2743	2375	0.1549	0.0007834

	t.pvalue	Sign.pvalue	Wilcox.pvalue
D_CC_LH_3_ICDF_0.99	0.002706	0.02541	0.00595
D_CC_HL_4_ICDF_0.999	0.01312	0.3177	0.01647
S_CC_GLCM_4_Inertia	0.0003231	0.0003834	0.0001619
S_CC_FRACTAL_ICDF_0.95	0.004744	0.002917	0.004736
D_CC_LH_3_ICDF_0.999	0.02278	0.1301	0.03753
S_CC_FRACTAL_ICDF_0.75	0.02969	0.09671	0.01799
S_CC_LH_2_Mean	0.07385	0.1681	0.06131
S_CC_FRACTAL_Mean	0.0283	0.06167	0.01766
S_CC_FRACTAL_ICDF_0.25	0.05441	0.112	0.03928

MSE: 1494
R2: 0.5183
bootstrap:

gain <- length(MODELFRESA$BSWiMS.models$formula.list)/bswimsReps
gplots::heatmap.2(gain*MODELFRESA$BSWiMS.models$bagging$formulaNetwork,trace="none",mar=c(10,10),main="B:SWiMS Formula Network")

Benchmarks for Recurence Risk

José Tamez-Peña

April 18, 2018

1 FRESA.CAD Regresion Benchmark

1.1 BRCA Recurence RISK Data Set

1.2 Benchmark

1.2.1 Results

1.3 Features Analysis