Examples of fitting models, predicting class labels, evaluating prediction, and analyzing fitting results

Description

These examples demonstrate how to use BCBCSF package. They use all prior and Markov chain sampling settings by default (except no_rmc as noted below). The methods for setting others can be found from documents for specific functions. However, the default settings may work well for a wide range of gene expression data.

References

Li, L. (2012), Bias-corrected Hierarchical Bayesian Classification with a Selected Subset of High-dimensional Features, Journal of American Statistical Association,107:497,120-134

See Also

bcbcsf_fitpred, bcbcsf_pred, cross_vld, eval_pred, reload_fit_bcbcsf, bcbcsf_sumfit, bcbcsf_plotsumfit

Examples

##\dontrun{
## load lymphoma microarray data
data (lymphoma)

## select some cases as testing data set
ts <- c (sort(sample (1:42,5)), 43:44, 61:62)

## training data
X_tr <- lymph.X[-ts,] 
y_tr <- lymph.y[-ts]
## test data 
X_ts <- lymph.X[ts,] 
y_ts <- lymph.y[ts] 

##########################################################################
######################## training and prediction #########################
##########################################################################
## fitting training data with top features selected by F-statistic 
out_fit <- bcbcsf_fitpred (X_tr = X_tr, y_tr = y_tr, nos_fsel = c(20, 50),
                           no_rmc = 100)
## note 1: if 'X_ts' is given above, prediction is made after fitting
## note 2: no_rmc = 100 is too small, omit it and use the default

## predicting class labels of test cases 
out_pred <- bcbcsf_pred (X_ts = X_ts, out_fit = out_fit)

## evaluate prediction given true labels
eval_pred (out_pred = out_pred, y_ts = y_ts)

##########################################################################
####################### visualizing prediction results ###################
##########################################################################
## reload one bcbcsf fit result from hardrive
fit_bcbcsf <- reload_fit_bcbcsf (out_fit$fitfiles[1])
## the fitting result for no_fsel = 50 can be retrieved directly from
## out_fit:
fit_bcbcsf_fsel50 <- out_fit$fit_bcbcsf
## summarize the fitting result
sum_fit <- bcbcsf_sumfit (fit_bcbcsf)

## visualize fitting result
bcbcsf_plotsumfit (sum_fit)

##########################################################################
############################ cross validation ############################
##########################################################################
## doing cross validation with bcbcsf_fitpred on lymphoma data 
cv_pred <- cross_vld (
       ##################### classifier, data, and fold ###################
       fitpred_func = bcbcsf_fitpred, X = lymph.X, y = lymph.y, nfold = 2,
       ################ all other arguments passed classifier ############
       nos_fsel = c(20,50), no_rmc = 100 )
## note: no_rmc = 100 is too small, omit it and use the default in practice

## evaluate prediction given true labels            
eval_pred (out_pred = cv_pred, y_ts = lymph.y)   

## warning: this function is slow if nfold is large; if you have a 
## computer cluster, you better parallel the cross validation folds.
##}

Functions for fitting models with MCMC, predicting class labels of test cases, and finding predictive probabilities with cross-validation

Description

bcbcsf_fitpred trains models with Gibbs sampling for each number of retained features. The results are saved in files. This function also makes predictions for test cases if they are provided.

bcbcsf_pred uses the posterior samples saved by bcbcsf_fitpred to predict the class labels of test cases. Prediction results are an array of predictive probabilities array_probs_pred, whose rows for test cases, columns for classes, and the 3rd dimension for different numbers of retained features.

cross_vld uses cross-validation to obtain predictive probabilities for all cases of a data set. This generic function can be used with bcbcsf_fitpred and other classifiers.

Usage

bcbcsf_fitpred (
  ## arguments specifying info of data sets
  X_tr, y_tr, nos_fsel = ncol (X_tr), 
  X_ts = NULL,  standardize = FALSE, rankf = FALSE,
  ## arguments for prediction
  burn = NULL, thin = 1, offset_sdxj = 0.5,
  ## arguments for Markov chain sampling
  no_rmc = 1000, no_imc = 5, no_mhwmux = 10,
  fit_bcbcsf_filepre = ".fitbcbcsf_", 
  ## arguments specifying priors for parameters and hyerparameters
  w0_mu = 0.05, alpha0_mu = 0.5, alpha1_mu = 3,
  w0_x  = 1.00, alpha0_x  = 0.5, alpha1_x  = 10,
  w0_nu = 0.05, alpha0_nu = 0.5, prior_psi = NULL,
  ## arguments for metropolis sampling for wmu, wx
  stepadj_mhwmux = 1, diag_mhwmux = FALSE,
  ## arguments for computing adjustment factor
  bcor = 1, cut_qf = exp (-10), cut_dpoi = exp (-10), nos_sim = 1000,
  ## whether look at progress
  monitor = TRUE)
  
bcbcsf_pred (X_ts, out_fit, burn = NULL, thin = 1, offset_sdxj = 0.5)

cross_vld (X, y, nfold = 10, folds = NULL, 
           fitpred_func = bcbcsf_fitpred,  ...)

Arguments

Value

A function for evaluating arrays of predictive probabilities with the true class labels of test cases

Description

This function is used to find error rate, amlp, loss and predictive probabilities at true labels.

Usage

eval_pred (out_pred, y_ts, Mloss = NULL)

Arguments

; by default, it is NULL.

Value

Functions for analyzing and visualizing a BCBCSF fitting result

Description

These functions are used to look at the fitting results, especially plot the gene signals.

Usage

reload_fit_bcbcsf (fit_bcbcsf_afile)

bcbcsf_sumfit (fit_bcbcsf = NULL, fit_bcbcsf_afile = NULL, 
               burn = NULL,  thin = 1)

bcbcsf_plotsumfit (sum_fit)
 

Arguments

Value

reload_fit_bcbcsf returns a list of Markov chain fitting results, including how to do feature selection and data preprocessing.

bcbcsf_sumfit returns a list of point estimates of means and variances.

bcbcsf_plotsumfit returns nothing; it plots the normalized means (for each gene, original expression means substracted by their means and divided by the common standard deviation), and overall signals (Euclid distance of normalized means) for the selected features.

Lymphoma Microarray Data

Description

This is one of the microarray data sets used to demonstrate BCBCSF in the reference article. Information about this data set can be found from the reference.

Usage

    data(lymphoma)

Value

Internal functions used in package ‘gausspred’

Description

This function is internal. Type the function name to look at its details