python类maximum()的实例源码

libscores.py 文件源码 项目:AutoML5 作者: djajetic 项目源码 文件源码 阅读 32 收藏 0 点赞 0 评论 0
def bac_metric (solution, prediction, task='binary.classification'):
    ''' Compute the normalized balanced accuracy. The binarization and 
    the normalization differ for the multi-label and multi-class case. '''
    label_num = solution.shape[1]
    score = np.zeros(label_num)
    bin_prediction = binarize_predictions(prediction, task)
    [tn,fp,tp,fn] = acc_stat(solution, bin_prediction)
    # Bounding to avoid division by 0
    eps = 1e-15
    tp = sp.maximum (eps, tp)
    pos_num = sp.maximum (eps, tp+fn)
    tpr = tp / pos_num # true positive rate (sensitivity)
    if (task != 'multiclass.classification') or (label_num==1):
        tn = sp.maximum (eps, tn)
        neg_num = sp.maximum (eps, tn+fp)
        tnr = tn / neg_num # true negative rate (specificity)
        bac = 0.5*(tpr + tnr)
        base_bac = 0.5     # random predictions for binary case
    else: 
        bac = tpr
        base_bac = 1./label_num # random predictions for multiclass case
    bac = mvmean(bac)     # average over all classes
    # Normalize: 0 for random, 1 for perfect
    score = (bac - base_bac) / sp.maximum(eps, (1 - base_bac))
    return score
libscores.py 文件源码 项目:AutoML4 作者: djajetic 项目源码 文件源码 阅读 31 收藏 0 点赞 0 评论 0
def bac_metric (solution, prediction, task='binary.classification'):
    ''' Compute the normalized balanced accuracy. The binarization and 
    the normalization differ for the multi-label and multi-class case. '''
    label_num = solution.shape[1]
    score = np.zeros(label_num)
    bin_prediction = binarize_predictions(prediction, task)
    [tn,fp,tp,fn] = acc_stat(solution, bin_prediction)
    # Bounding to avoid division by 0
    eps = 1e-15
    tp = sp.maximum (eps, tp)
    pos_num = sp.maximum (eps, tp+fn)
    tpr = tp / pos_num # true positive rate (sensitivity)
    if (task != 'multiclass.classification') or (label_num==1):
        tn = sp.maximum (eps, tn)
        neg_num = sp.maximum (eps, tn+fp)
        tnr = tn / neg_num # true negative rate (specificity)
        bac = 0.5*(tpr + tnr)
        base_bac = 0.5     # random predictions for binary case
    else: 
        bac = tpr
        base_bac = 1./label_num # random predictions for multiclass case
    bac = mvmean(bac)     # average over all classes
    # Normalize: 0 for random, 1 for perfect
    score = (bac - base_bac) / sp.maximum(eps, (1 - base_bac))
    return score
libscores.py 文件源码 项目:AutoML4 作者: djajetic 项目源码 文件源码 阅读 43 收藏 0 点赞 0 评论 0
def prior_log_loss(frac_pos, task = 'binary.classification'):
    ''' Baseline log loss. For multiplr classes ot labels return the volues for each column'''
    eps = 1e-15   
    frac_pos_ = sp.maximum (eps, frac_pos)
    if (task != 'multiclass.classification'): # binary case
        frac_neg = 1-frac_pos
        frac_neg_ = sp.maximum (eps, frac_neg)
        pos_class_log_loss_ = - frac_pos * np.log(frac_pos_)
        neg_class_log_loss_ = - frac_neg * np.log(frac_neg_)
        base_log_loss = pos_class_log_loss_ + neg_class_log_loss_
        # base_log_loss = mvmean(base_log_loss)
        # print('binary {}'.format(base_log_loss))
        # In the multilabel case, the right thing i to AVERAGE not sum
        # We return all the scores so we can normalize correctly later on
    else: # multiclass case
        fp = frac_pos_ / sum(frac_pos_) # Need to renormalize the lines in multiclass case
        # Only ONE label is 1 in the multiclass case active for each line
        pos_class_log_loss_ = - frac_pos * np.log(fp)
        base_log_loss = np.sum(pos_class_log_loss_) 
    return base_log_loss

# sklearn implementations for comparison
libscores.py 文件源码 项目:automl_gpu 作者: abhishekkrthakur 项目源码 文件源码 阅读 27 收藏 0 点赞 0 评论 0
def bac_metric (solution, prediction, task='binary.classification'):
    ''' Compute the normalized balanced accuracy. The binarization and 
    the normalization differ for the multi-label and multi-class case. '''
    label_num = solution.shape[1]
    score = np.zeros(label_num)
    bin_prediction = binarize_predictions(prediction, task)
    [tn,fp,tp,fn] = acc_stat(solution, bin_prediction)
    # Bounding to avoid division by 0
    eps = 1e-15
    tp = sp.maximum (eps, tp)
    pos_num = sp.maximum (eps, tp+fn)
    tpr = tp / pos_num # true positive rate (sensitivity)
    if (task != 'multiclass.classification') or (label_num==1):
        tn = sp.maximum (eps, tn)
        neg_num = sp.maximum (eps, tn+fp)
        tnr = tn / neg_num # true negative rate (specificity)
        bac = 0.5*(tpr + tnr)
        base_bac = 0.5     # random predictions for binary case
    else: 
        bac = tpr
        base_bac = 1./label_num # random predictions for multiclass case
    bac = mvmean(bac)     # average over all classes
    # Normalize: 0 for random, 1 for perfect
    score = (bac - base_bac) / sp.maximum(eps, (1 - base_bac))
    return score
libscores.py 文件源码 项目:automl_gpu 作者: abhishekkrthakur 项目源码 文件源码 阅读 37 收藏 0 点赞 0 评论 0
def prior_log_loss(frac_pos, task = 'binary.classification'):
    ''' Baseline log loss. For multiplr classes ot labels return the volues for each column'''
    eps = 1e-15   
    frac_pos_ = sp.maximum (eps, frac_pos)
    if (task != 'multiclass.classification'): # binary case
        frac_neg = 1-frac_pos
        frac_neg_ = sp.maximum (eps, frac_neg)
        pos_class_log_loss_ = - frac_pos * np.log(frac_pos_)
        neg_class_log_loss_ = - frac_neg * np.log(frac_neg_)
        base_log_loss = pos_class_log_loss_ + neg_class_log_loss_
        # base_log_loss = mvmean(base_log_loss)
        # print('binary {}'.format(base_log_loss))
        # In the multilabel case, the right thing i to AVERAGE not sum
        # We return all the scores so we can normalize correctly later on
    else: # multiclass case
        fp = frac_pos_ / sum(frac_pos_) # Need to renormalize the lines in multiclass case
        # Only ONE label is 1 in the multiclass case active for each line
        pos_class_log_loss_ = - frac_pos * np.log(fp)
        base_log_loss = np.sum(pos_class_log_loss_) 
    return base_log_loss

# sklearn implementations for comparison
libscores.py 文件源码 项目:AutoML5 作者: djajetic 项目源码 文件源码 阅读 33 收藏 0 点赞 0 评论 0
def pac_metric (solution, prediction, task='binary.classification'):
    ''' Probabilistic Accuracy based on log_loss metric. 
    We assume the solution is in {0, 1} and prediction in [0, 1].
    Otherwise, run normalize_array.''' 
    debug_flag=False
    [sample_num, label_num] = solution.shape
    if label_num==1: task='binary.classification'
    eps = 1e-15
    the_log_loss = log_loss(solution, prediction, task)
    # Compute the base log loss (using the prior probabilities)    
    pos_num = 1.* sum(solution) # float conversion!
    frac_pos = pos_num / sample_num # prior proba of positive class
    the_base_log_loss = prior_log_loss(frac_pos, task)
    # Alternative computation of the same thing (slower)    
    # Should always return the same thing except in the multi-label case
    # For which the analytic solution makes more sense
    if debug_flag:
        base_prediction = np.empty(prediction.shape)
        for k in range(sample_num): base_prediction[k,:] = frac_pos
        base_log_loss = log_loss(solution, base_prediction, task)  
        diff = np.array(abs(the_base_log_loss-base_log_loss))
        if len(diff.shape)>0: diff=max(diff)
        if(diff)>1e-10: 
            print('Arrggh {} != {}'.format(the_base_log_loss,base_log_loss))
    # Exponentiate to turn into an accuracy-like score.
    # In the multi-label case, we need to average AFTER taking the exp 
    # because it is an NL operation
    pac = mvmean(np.exp(-the_log_loss)) 
    base_pac = mvmean(np.exp(-the_base_log_loss))
    # Normalize: 0 for random, 1 for perfect    
    score = (pac - base_pac) / sp.maximum(eps, (1 - base_pac))
    return score
libscores.py 文件源码 项目:AutoML5 作者: djajetic 项目源码 文件源码 阅读 33 收藏 0 点赞 0 评论 0
def log_loss(solution, prediction, task = 'binary.classification'):
    ''' Log loss for binary and multiclass. '''
    [sample_num, label_num] = solution.shape
    eps = 1e-15

    pred = np.copy(prediction) # beware: changes in prediction occur through this
    sol = np.copy(solution)
    if (task == 'multiclass.classification') and (label_num>1):
        # Make sure the lines add up to one for multi-class classification
        norma = np.sum(prediction, axis=1)
        for k in range(sample_num):
            pred[k,:] /= sp.maximum (norma[k], eps) 
        # Make sure there is a single label active per line for multi-class classification
        sol = binarize_predictions(solution, task='multiclass.classification')
        # For the base prediction, this solution is ridiculous in the multi-label case

    # Bounding of predictions to avoid log(0),1/0,...
    pred = sp.minimum (1-eps, sp.maximum (eps, pred))
    # Compute the log loss    
    pos_class_log_loss = - mvmean(sol*np.log(pred), axis=0)
    if (task != 'multiclass.classification') or (label_num==1):
        # The multi-label case is a bunch of binary problems.
        # The second class is the negative class for each column.
        neg_class_log_loss = - mvmean((1-sol)*np.log(1-pred), axis=0)
        log_loss = pos_class_log_loss + neg_class_log_loss
        # Each column is an independent problem, so we average.
        # The probabilities in one line do not add up to one.
        # log_loss = mvmean(log_loss) 
        # print('binary {}'.format(log_loss))
        # In the multilabel case, the right thing i to AVERAGE not sum
        # We return all the scores so we can normalize correctly later on
    else:
        # For the multiclass case the probabilities in one line add up one.
        log_loss = pos_class_log_loss
        # We sum the contributions of the columns.
        log_loss = np.sum(log_loss) 
        #print('multiclass {}'.format(log_loss))
    return log_loss
libscores.py 文件源码 项目:AutoML5 作者: djajetic 项目源码 文件源码 阅读 37 收藏 0 点赞 0 评论 0
def prior_log_loss(frac_pos, task = 'binary.classification'):
    ''' Baseline log loss. For multiplr classes ot labels return the volues for each column'''
    eps = 1e-15   
    frac_pos_ = sp.maximum (eps, frac_pos)
    if (task != 'multiclass.classification'): # binary case
        frac_neg = 1-frac_pos
        frac_neg_ = sp.maximum (eps, frac_neg)
        pos_class_log_loss_ = - frac_pos * np.log(frac_pos_)
        neg_class_log_loss_ = - frac_neg * np.log(frac_neg_)
        base_log_loss = pos_class_log_loss_ + neg_class_log_loss_
        # base_log_loss = mvmean(base_log_loss)
        # print('binary {}'.format(base_log_loss))
        # In the multilabel case, the right thing i to AVERAGE not sum
        # We return all the scores so we can normalize correctly later on
    else: # multiclass case
        fp = frac_pos_ / sum(frac_pos_) # Need to renormalize the lines in multiclass case
        # Only ONE label is 1 in the multiclass case active for each line
        pos_class_log_loss_ = - frac_pos * np.log(fp)
        base_log_loss = np.sum(pos_class_log_loss_) 
    return base_log_loss

# sklearn implementations for comparison
np_utils.py 文件源码 项目:CopyNet 作者: MultiPath 项目源码 文件源码 阅读 41 收藏 0 点赞 0 评论 0
def binary_logloss(p, y):
    epsilon = 1e-15
    p = sp.maximum(epsilon, p)
    p = sp.minimum(1-epsilon, p)
    res = sum(y * sp.log(p) + sp.subtract(1, y) * sp.log(sp.subtract(1, p)))
    res *= -1.0/len(y)
    return res
libscores.py 文件源码 项目:AutoML4 作者: djajetic 项目源码 文件源码 阅读 29 收藏 0 点赞 0 评论 0
def pac_metric (solution, prediction, task='binary.classification'):
    ''' Probabilistic Accuracy based on log_loss metric. 
    We assume the solution is in {0, 1} and prediction in [0, 1].
    Otherwise, run normalize_array.''' 
    debug_flag=False
    [sample_num, label_num] = solution.shape
    if label_num==1: task='binary.classification'
    eps = 1e-15
    the_log_loss = log_loss(solution, prediction, task)
    # Compute the base log loss (using the prior probabilities)    
    pos_num = 1.* sum(solution) # float conversion!
    frac_pos = pos_num / sample_num # prior proba of positive class
    the_base_log_loss = prior_log_loss(frac_pos, task)
    # Alternative computation of the same thing (slower)    
    # Should always return the same thing except in the multi-label case
    # For which the analytic solution makes more sense
    if debug_flag:
        base_prediction = np.empty(prediction.shape)
        for k in range(sample_num): base_prediction[k,:] = frac_pos
        base_log_loss = log_loss(solution, base_prediction, task)  
        diff = np.array(abs(the_base_log_loss-base_log_loss))
        if len(diff.shape)>0: diff=max(diff)
        if(diff)>1e-10: 
            print('Arrggh {} != {}'.format(the_base_log_loss,base_log_loss))
    # Exponentiate to turn into an accuracy-like score.
    # In the multi-label case, we need to average AFTER taking the exp 
    # because it is an NL operation
    pac = mvmean(np.exp(-the_log_loss)) 
    base_pac = mvmean(np.exp(-the_base_log_loss))
    # Normalize: 0 for random, 1 for perfect    
    score = (pac - base_pac) / sp.maximum(eps, (1 - base_pac))
    return score
libscores.py 文件源码 项目:AutoML4 作者: djajetic 项目源码 文件源码 阅读 30 收藏 0 点赞 0 评论 0
def log_loss(solution, prediction, task = 'binary.classification'):
    ''' Log loss for binary and multiclass. '''
    [sample_num, label_num] = solution.shape
    eps = 1e-15

    pred = np.copy(prediction) # beware: changes in prediction occur through this
    sol = np.copy(solution)
    if (task == 'multiclass.classification') and (label_num>1):
        # Make sure the lines add up to one for multi-class classification
        norma = np.sum(prediction, axis=1)
        for k in range(sample_num):
            pred[k,:] /= sp.maximum (norma[k], eps) 
        # Make sure there is a single label active per line for multi-class classification
        sol = binarize_predictions(solution, task='multiclass.classification')
        # For the base prediction, this solution is ridiculous in the multi-label case

    # Bounding of predictions to avoid log(0),1/0,...
    pred = sp.minimum (1-eps, sp.maximum (eps, pred))
    # Compute the log loss    
    pos_class_log_loss = - mvmean(sol*np.log(pred), axis=0)
    if (task != 'multiclass.classification') or (label_num==1):
        # The multi-label case is a bunch of binary problems.
        # The second class is the negative class for each column.
        neg_class_log_loss = - mvmean((1-sol)*np.log(1-pred), axis=0)
        log_loss = pos_class_log_loss + neg_class_log_loss
        # Each column is an independent problem, so we average.
        # The probabilities in one line do not add up to one.
        # log_loss = mvmean(log_loss) 
        # print('binary {}'.format(log_loss))
        # In the multilabel case, the right thing i to AVERAGE not sum
        # We return all the scores so we can normalize correctly later on
    else:
        # For the multiclass case the probabilities in one line add up one.
        log_loss = pos_class_log_loss
        # We sum the contributions of the columns.
        log_loss = np.sum(log_loss) 
        #print('multiclass {}'.format(log_loss))
    return log_loss
beacon_analysis.py 文件源码 项目:digital_rf 作者: MITHaystack 项目源码 文件源码 阅读 33 收藏 0 点赞 0 评论 0
def open_file(maindir):
    """
        Creates the digital RF reading object.

        Args:
            maindir (:obj:'str'): The directory where the data is located.

        Returns:
            drfObj (obj:"DigitalRFReader"): Digital RF Reader object.
            chandict (obj:"dict"): Dictionary that holds info for the channels.
            start_indx (obj:'long'): Start index in samples.
            end_indx (obj:'long'): End index in samples.
    """
    mainpath = os.path.expanduser(maindir)
    drfObj = drf.DigitalRFReader(mainpath)
    chans = drfObj.get_channels()

    chandict={}

    start_indx, end_indx=[0, sp.inf]
    # Get channel info
    for ichan in chans:
        curdict = {}

        curdict['sind'], curdict['eind'] = drfObj.get_bounds(ichan)
        # determine the read boundrys assuming the sampling is the same.
        start_indx = sp.maximum(curdict['sind'], start_indx)
        end_indx = sp.minimum(curdict['eind'], end_indx)

        dmetadict = drfObj.read_metadata(start_indx, end_indx, ichan)
        dmetakeys = dmetadict.keys()

        curdict['sps'] = dmetadict[dmetakeys[0]]['samples_per_second']
        curdict['fo'] = dmetadict[dmetakeys[0]]['center_frequencies'].ravel()[0]
        chandict[ichan] = curdict

    return (drfObj, chandict, start_indx, end_indx)
np_utils.py 文件源码 项目:keras 作者: GeekLiB 项目源码 文件源码 阅读 25 收藏 0 点赞 0 评论 0
def binary_logloss(p, y):
    epsilon = 1e-15
    p = sp.maximum(epsilon, p)
    p = sp.minimum(1-epsilon, p)
    res = sum(y * sp.log(p) + sp.subtract(1, y) * sp.log(sp.subtract(1, p)))
    res *= -1.0/len(y)
    return res
libscores.py 文件源码 项目:automl_gpu 作者: abhishekkrthakur 项目源码 文件源码 阅读 26 收藏 0 点赞 0 评论 0
def pac_metric (solution, prediction, task='binary.classification'):
    ''' Probabilistic Accuracy based on log_loss metric. 
    We assume the solution is in {0, 1} and prediction in [0, 1].
    Otherwise, run normalize_array.''' 
    debug_flag=False
    [sample_num, label_num] = solution.shape
    if label_num==1: task='binary.classification'
    eps = 1e-15
    the_log_loss = log_loss(solution, prediction, task)
    # Compute the base log loss (using the prior probabilities)    
    pos_num = 1.* sum(solution) # float conversion!
    frac_pos = pos_num / sample_num # prior proba of positive class
    the_base_log_loss = prior_log_loss(frac_pos, task)
    # Alternative computation of the same thing (slower)    
    # Should always return the same thing except in the multi-label case
    # For which the analytic solution makes more sense
    if debug_flag:
        base_prediction = np.empty(prediction.shape)
        for k in range(sample_num): base_prediction[k,:] = frac_pos
        base_log_loss = log_loss(solution, base_prediction, task)  
        diff = np.array(abs(the_base_log_loss-base_log_loss))
        if len(diff.shape)>0: diff=max(diff)
        if(diff)>1e-10: 
            print('Arrggh {} != {}'.format(the_base_log_loss,base_log_loss))
    # Exponentiate to turn into an accuracy-like score.
    # In the multi-label case, we need to average AFTER taking the exp 
    # because it is an NL operation
    pac = mvmean(np.exp(-the_log_loss)) 
    base_pac = mvmean(np.exp(-the_base_log_loss))
    # Normalize: 0 for random, 1 for perfect    
    score = (pac - base_pac) / sp.maximum(eps, (1 - base_pac))
    return score
libscores.py 文件源码 项目:automl_gpu 作者: abhishekkrthakur 项目源码 文件源码 阅读 31 收藏 0 点赞 0 评论 0
def log_loss(solution, prediction, task = 'binary.classification'):
    ''' Log loss for binary and multiclass. '''
    [sample_num, label_num] = solution.shape
    eps = 1e-15

    pred = np.copy(prediction) # beware: changes in prediction occur through this
    sol = np.copy(solution)
    if (task == 'multiclass.classification') and (label_num>1):
        # Make sure the lines add up to one for multi-class classification
        norma = np.sum(prediction, axis=1)
        for k in range(sample_num):
            pred[k,:] /= sp.maximum (norma[k], eps) 
        # Make sure there is a single label active per line for multi-class classification
        sol = binarize_predictions(solution, task='multiclass.classification')
        # For the base prediction, this solution is ridiculous in the multi-label case

    # Bounding of predictions to avoid log(0),1/0,...
    pred = sp.minimum (1-eps, sp.maximum (eps, pred))
    # Compute the log loss    
    pos_class_log_loss = - mvmean(sol*np.log(pred), axis=0)
    if (task != 'multiclass.classification') or (label_num==1):
        # The multi-label case is a bunch of binary problems.
        # The second class is the negative class for each column.
        neg_class_log_loss = - mvmean((1-sol)*np.log(1-pred), axis=0)
        log_loss = pos_class_log_loss + neg_class_log_loss
        # Each column is an independent problem, so we average.
        # The probabilities in one line do not add up to one.
        # log_loss = mvmean(log_loss) 
        # print('binary {}'.format(log_loss))
        # In the multilabel case, the right thing i to AVERAGE not sum
        # We return all the scores so we can normalize correctly later on
    else:
        # For the multiclass case the probabilities in one line add up one.
        log_loss = pos_class_log_loss
        # We sum the contributions of the columns.
        log_loss = np.sum(log_loss) 
        #print('multiclass {}'.format(log_loss))
    return log_loss
final.py 文件源码 项目:pCVR 作者: xjtushilei 项目源码 文件源码 阅读 37 收藏 0 点赞 0 评论 0
def loglossl(act, pred):
    epsilon = 1e-15
    pred = sp.maximum(epsilon, pred)
    pred = sp.minimum(1-epsilon, pred)
    ll = sum(act*sp.log(pred) + sp.subtract(1,act)*sp.log(sp.subtract(1,pred)))
    ll = ll * -1.0/len(act)
    return ll
utils.py 文件源码 项目:pCVR 作者: xjtushilei 项目源码 文件源码 阅读 25 收藏 0 点赞 0 评论 0
def my_logloss(act, pred):
    epsilon = 1e-15
    pred = K.maximum(epsilon, pred)
    pred = K.minimum(1 - epsilon, pred)
    ll = K.sum(act * K.log(pred) + (1 - act) * K.log(1 - pred))
    ll = ll * -1.0 / K.shape(act)[0]

    return ll
utils.py 文件源码 项目:pCVR 作者: xjtushilei 项目源码 文件源码 阅读 35 收藏 0 点赞 0 评论 0
def logloss(act, pred):
    '''
    ????????
    :param act: 
    :param pred: 
    :return: 
    '''
    epsilon = 1e-15
    pred = sp.maximum(epsilon, pred)
    pred = sp.minimum(1 - epsilon, pred)
    ll = sum(act * sp.log(pred) + sp.subtract(1, act) * sp.log(sp.subtract(1, pred)))
    ll = ll * -1.0 / len(act)
    return ll
np_utils.py 文件源码 项目:seq2seq-keyphrase 作者: memray 项目源码 文件源码 阅读 30 收藏 0 点赞 0 评论 0
def binary_logloss(p, y):
    epsilon = 1e-15
    p = sp.maximum(epsilon, p)
    p = sp.minimum(1-epsilon, p)
    res = sum(y * sp.log(p) + sp.subtract(1, y) * sp.log(sp.subtract(1, p)))
    res *= -1.0/len(y)
    return res
classification_metrics.py 文件源码 项目:AutoML-Challenge 作者: postech-mlg-exbrain 项目源码 文件源码 阅读 27 收藏 0 点赞 0 评论 0
def acc_metric(solution, prediction, task=BINARY_CLASSIFICATION):
    """
    Compute the accuracy.

    Get the accuracy stats
    acc = (tpr + fpr) / (tn + fp + tp + fn)
    Normalize, so 1 is the best and zero mean random...

    :param solution:
    :param prediction:
    :param task:
    :return:
    """

    label_num = solution.shape[1]
    bin_predictions = binarize_predictions(prediction, task)
    tn, fp, tp, fn = acc_stat(solution, bin_predictions)
    # Bounding to avoid division by 0
    eps = np.float(1e-15)
    tp = np.sum(tp)
    fp = np.sum(fp)
    tn = np.sum(tn)
    fn = np.sum(fn)

    if (task != MULTICLASS_CLASSIFICATION) or (label_num == 1):
        accuracy = (np.sum(tp) + np.sum(tn)) / (
            np.sum(tp) + np.sum(fp) + np.sum(tn) + np.sum(fn)
        )
    else:
        accuracy = np.sum(tp) / (np.sum(tp) + np.sum(fp))

    if (task != MULTICLASS_CLASSIFICATION) or (label_num == 1):
        base_accuracy = 0.5  # random predictions for binary case
    else:
        base_accuracy = 1. / label_num
    # Normalize: 0 for random, 1 for perfect
    score = (accuracy - base_accuracy) / sp.maximum(eps, (1 - base_accuracy))
    return score
classification_metrics.py 文件源码 项目:AutoML-Challenge 作者: postech-mlg-exbrain 项目源码 文件源码 阅读 28 收藏 0 点赞 0 评论 0
def bac_metric(solution, prediction, task=BINARY_CLASSIFICATION):
    """
    Compute the normalized balanced accuracy.

    The binarization and
    the normalization differ for the multi-label and multi-class case.
    :param solution:
    :param prediction:
    :param task:
    :return:
    """

    label_num = solution.shape[1]
    score = np.zeros(label_num)
    bin_prediction = binarize_predictions(prediction, task)
    [tn, fp, tp, fn] = acc_stat(solution, bin_prediction)
    # Bounding to avoid division by 0
    eps = 1e-15
    tp = sp.maximum(eps, tp)
    pos_num = sp.maximum(eps, tp + fn)
    tpr = tp / pos_num  # true positive rate (sensitivity)
    if (task != MULTICLASS_CLASSIFICATION) or (label_num == 1):
        tn = sp.maximum(eps, tn)
        neg_num = sp.maximum(eps, tn + fp)
        tnr = tn / neg_num  # true negative rate (specificity)
        bac = 0.5 * (tpr + tnr)
        base_bac = 0.5  # random predictions for binary case
    else:
        bac = tpr
        base_bac = 1. / label_num  # random predictions for multiclass case
    bac = np.mean(bac)  # average over all classes
    # Normalize: 0 for random, 1 for perfect
    score = (bac - base_bac) / sp.maximum(eps, (1 - base_bac))
    return score
util.py 文件源码 项目:AutoML-Challenge 作者: postech-mlg-exbrain 项目源码 文件源码 阅读 43 收藏 0 点赞 0 评论 0
def log_loss(solution, prediction, task=BINARY_CLASSIFICATION):
    """Log loss for binary and multiclass."""
    [sample_num, label_num] = solution.shape
    eps = 1e-15

    pred = np.copy(prediction)  # beware: changes in prediction occur through this
    sol = np.copy(solution)
    if (task == MULTICLASS_CLASSIFICATION) and (label_num > 1):
        # Make sure the lines add up to one for multi-class classification
        norma = np.sum(prediction, axis=1)
        for k in range(sample_num):
            pred[k, :] /= sp.maximum(norma[k], eps)
        # Make sure there is a single label active per line for multi-class
        # classification
        sol = binarize_predictions(solution, task=MULTICLASS_CLASSIFICATION)
        # For the base prediction, this solution is ridiculous in the
        # multi-label case

        # Bounding of predictions to avoid log(0),1/0,...
    pred = sp.minimum(1 - eps, sp.maximum(eps, pred))
    # Compute the log loss
    pos_class_log_loss = -np.mean(sol * np.log(pred), axis=0)
    if (task != MULTICLASS_CLASSIFICATION) or (label_num == 1):
        # The multi-label case is a bunch of binary problems.
        # The second class is the negative class for each column.
        neg_class_log_loss = -np.mean((1 - sol) * np.log(1 - pred), axis=0)
        log_loss = pos_class_log_loss + neg_class_log_loss
        # Each column is an independent problem, so we average.
        # The probabilities in one line do not add up to one.
        # log_loss = mvmean(log_loss)
        # print('binary {}'.format(log_loss))
        # In the multilabel case, the right thing i to AVERAGE not sum
        # We return all the scores so we can normalize correctly later on
    else:
        # For the multiclass case the probabilities in one line add up one.
        log_loss = pos_class_log_loss
        # We sum the contributions of the columns.
        log_loss = np.sum(log_loss)
        # print('multiclass {}'.format(log_loss))
    return log_loss
np_utils.py 文件源码 项目:keraflow 作者: ipod825 项目源码 文件源码 阅读 47 收藏 0 点赞 0 评论 0
def binary_logloss(p, y):
    epsilon = 1e-15
    p = sp.maximum(epsilon, p)
    p = sp.minimum(1-epsilon, p)
    res = sum(y * sp.log(p) + sp.subtract(1, y) * sp.log(sp.subtract(1, p)))
    res *= -1.0/len(y)
    return res
XGBoostTest.py 文件源码 项目:Tencent2017_Final_Coda_Allegro 作者: BladeCoda 项目源码 文件源码 阅读 39 收藏 0 点赞 0 评论 0
def logloss(act, pred):
  epsilon = 1e-15
  pred = sp.maximum(epsilon, pred)
  pred = sp.minimum(1-epsilon, pred)
  ll = sum(act*sp.log(pred) + sp.subtract(1,act)*sp.log(sp.subtract(1,pred)))
  ll = ll * -1.0/len(act)
  return ll
LightGBMExam.py 文件源码 项目:Tencent2017_Final_Coda_Allegro 作者: BladeCoda 项目源码 文件源码 阅读 29 收藏 0 点赞 0 评论 0
def logloss(act, pred):
  epsilon = 1e-15
  pred = sp.maximum(epsilon, pred)
  pred = sp.minimum(1-epsilon, pred)
  ll = sum(act*sp.log(pred) + sp.subtract(1,act)*sp.log(sp.subtract(1,pred)))
  ll = ll * -1.0/len(act)
  return ll
np_utils.py 文件源码 项目:keras-recommendation 作者: sonyisme 项目源码 文件源码 阅读 41 收藏 0 点赞 0 评论 0
def binary_logloss(p, y):
    epsilon = 1e-15
    p = sp.maximum(epsilon, p)
    p = sp.minimum(1-epsilon, p)
    res = sum(y*sp.log(p) + sp.subtract(1,y)*sp.log(sp.subtract(1,p)))
    res *= -1.0/len(y)
    return res
np_utils.py 文件源码 项目:keras-customized 作者: ambrite 项目源码 文件源码 阅读 38 收藏 0 点赞 0 评论 0
def binary_logloss(p, y):
    epsilon = 1e-15
    p = sp.maximum(epsilon, p)
    p = sp.minimum(1-epsilon, p)
    res = sum(y * sp.log(p) + sp.subtract(1, y) * sp.log(sp.subtract(1, p)))
    res *= -1.0/len(y)
    return res
doFeats_1.py 文件源码 项目:Tencent_Social_Ads 作者: freelzy 项目源码 文件源码 阅读 31 收藏 0 点赞 0 评论 0
def logloss(act, preds):
    epsilon = 1e-15
    preds = sp.maximum(epsilon, preds)
    preds = sp.minimum(1 - epsilon, preds)
    ll = sum(act * sp.log(preds) + sp.subtract(1, act) * sp.log(sp.subtract(1, preds)))
    ll = ll * -1.0 / len(act)
    return ll
train_cv.py 文件源码 项目:Tencent_Social_Ads 作者: freelzy 项目源码 文件源码 阅读 24 收藏 0 点赞 0 评论 0
def logloss(act, preds):
    epsilon = 1e-15
    preds = sp.maximum(epsilon, preds)
    preds = sp.minimum(1 - epsilon, preds)
    ll = sum(act * sp.log(preds) + sp.subtract(1, act) * sp.log(sp.subtract(1, preds)))
    ll = ll * -1.0 / len(act)
    return ll
doFeats_2.py 文件源码 项目:Tencent_Social_Ads 作者: freelzy 项目源码 文件源码 阅读 28 收藏 0 点赞 0 评论 0
def logloss(act, preds):
    epsilon = 1e-15
    preds = sp.maximum(epsilon, preds)
    preds = sp.minimum(1 - epsilon, preds)
    ll = sum(act * sp.log(preds) + sp.subtract(1, act) * sp.log(sp.subtract(1, preds)))
    ll = ll * -1.0 / len(act)
    return ll


问题


面经


文章

微信
公众号

扫码关注公众号