'alphatau': alphatau, 'betatau': betatau, 'alpha0': alpha0, 'beta0': beta0, 'lambdaU': lambdaU, 'lambdaV': lambdaV } ''' Load in data. ''' R, M = load_gdsc_ic50() I, J = M.shape ''' Generate matrices M - one list of M's for each value of K. ''' M_attempts = 1000 all_Ms_training_and_test = [ compute_folds_attempts(I=I, J=J, no_folds=no_folds, attempts=M_attempts, M=M) for K in values_K ] ''' We now run the Gibbs sampler on each of the M's for each fraction. ''' all_performances = {metric: [] for metric in metrics} average_performances = {metric: [] for metric in metrics} # averaged over repeats for K, (Ms_train, Ms_test) in zip(values_K, all_Ms_training_and_test): print "Trying K=%s." % K # Run the algorithm <repeats> times and store all the performances for metric in metrics: all_performances[metric].append([]) for fold, (M_train, M_test) in enumerate(zip(Ms_train, Ms_test)): print "Fold %s of K=%s." % (fold + 1, K)
lambdaF, lambdaS, lambdaG = 0.1, 0.1, 0.1 alphatau, betatau = 1., 1. alpha0, beta0 = 1., 1. hyperparams = { 'alphatau':alphatau, 'betatau':betatau, 'alpha0':alpha0, 'beta0':beta0, 'lambdaF':lambdaF, 'lambdaS':lambdaS, 'lambdaG':lambdaG } ''' Load in data. ''' R, M = load_ccle_ec50() I, J = M.shape ''' Generate matrices M - one list of M's for each value of K. ''' M_attempts = 1000 all_Ms_training_and_test = [ compute_folds_attempts(I=I,J=J,no_folds=no_folds,attempts=M_attempts,M=M) for KL in values_KL ] ''' We now run the Gibbs sampler on each of the M's for each fraction. ''' all_performances = {metric:[] for metric in metrics} average_performances = {metric:[] for metric in metrics} # averaged over repeats for KL,(Ms_train,Ms_test) in zip(values_KL,all_Ms_training_and_test): print "Trying K,L=%s." % KL # Run the algorithm <repeats> times and store all the performances for metric in metrics: all_performances[metric].append([]) for fold,(M_train,M_test) in enumerate(zip(Ms_train,Ms_test)): print "Fold %s of K,L=%s." % (fold+1, KL)