comparison Marea/marea_cluster.py @ 34:1a97d1537623 draft

Lot of bug fixes

33:abf0bfe01c78

            labels = all_labels[i]
            predict = [x+1 for x in labels]
            classe = (pd.DataFrame(list(zip(dataset.index, predict)))).astype(str)
            classe.to_csv(best_cluster, sep = '\t', index = False, header = ['Patient_ID', 'Class'])
            
            
        if davies:
            with np.errstate(divide='ignore', invalid='ignore'):
                davies_bouldin = davies_bouldin_score(dataset, all_labels[i])
            warning("\nFor n_clusters = " + str(i + k_min) +
                  " The average davies bouldin score is: " + str(davies_bouldin))
        
       
        if silhouette:
            silihouette_draw(dataset, all_labels[i], i + k_min, 'clustering/silhouette_with_' + str(i + k_min) + prefix + '_clusters.png')
        

    # Number of clusters in labels, ignoring noise if present.
    n_clusters_ = len(set(labels)) - (1 if -1 in labels else 0)
    
    
    ##TODO: PLOT SU DBSCAN (no centers) e HIERARCHICAL
    
    labels = labels
    predict = [x+1 for x in labels]
    classe = (pd.DataFrame(list(zip(dataset.index, predict)))).astype(str)
    classe.to_csv(best_cluster, sep = '\t', index = False, header = ['Patient_ID', 'Class'])
  
    
########################## hierachical #######################################
    
def hierachical_agglomerative(dataset, k_min, k_max, best_cluster):

    if not os.path.exists('clustering'):
        os.makedirs('clustering')
    
    plt.figure(figsize=(10, 7))  

    
    range_n_clusters = [i for i in range(k_min, k_max+1)]

    scores = []
    labels = []
    for n_clusters in range_n_clusters:    
        cluster = AgglomerativeClustering(n_clusters=n_clusters, affinity='euclidean', linkage='ward')  
        cluster.fit_predict(dataset)  
        cluster_labels = cluster.labels_
        labels.append(cluster_labels)
        silhouette_avg = silhouette_score(dataset, cluster_labels)
        write_to_csv(dataset, cluster_labels, 'clustering/hierarchical_with_' + str(n_clusters) + '_clusters.tsv')
        scores.append(silhouette_avg)
        #warning("For n_clusters =", n_clusters,
              #"The average silhouette_score is :", silhouette_avg)
              
    best = max_index(scores) + k_min
     
    for i in range(len(labels)):
        if (i + k_min == best):
            labels = labels[i]
            predict = [x+1 for x in labels]
            classe = (pd.DataFrame(list(zip(dataset.index, predict)))).astype(str)
            classe.to_csv(best_cluster, sep = '\t', index = False, header = ['Patient_ID', 'Class'])
     
        
        
       

    
############################# main ###########################################


def main():

    
    if args.cluster_type == 'dbscan':
        dbscan(X, args.eps, args.min_samples, args.best_cluster)
        
    if args.cluster_type == 'hierarchy':
        hierachical_agglomerative(X, args.k_min, args.k_max, args.best_cluster)
        
##############################################################################

if __name__ == "__main__":
    main()

34:1a97d1537623

            labels = all_labels[i]
            predict = [x+1 for x in labels]
            classe = (pd.DataFrame(list(zip(dataset.index, predict)))).astype(str)
            classe.to_csv(best_cluster, sep = '\t', index = False, header = ['Patient_ID', 'Class'])
            
          
        
       
        if silhouette:
            silihouette_draw(dataset, all_labels[i], i + k_min, 'clustering/silhouette_with_' + str(i + k_min) + prefix + '_clusters.png')
        

    # Number of clusters in labels, ignoring noise if present.
    n_clusters_ = len(set(labels)) - (1 if -1 in labels else 0)
    
    
    labels = labels
    predict = [x+1 for x in labels]
    classe = (pd.DataFrame(list(zip(dataset.index, predict)))).astype(str)
    classe.to_csv(best_cluster, sep = '\t', index = False, header = ['Patient_ID', 'Class'])
  
    
########################## hierachical #######################################
    
def hierachical_agglomerative(dataset, k_min, k_max, best_cluster, silhouette):

    if not os.path.exists('clustering'):
        os.makedirs('clustering')
    
    plt.figure(figsize=(10, 7))  

    
    range_n_clusters = [i for i in range(k_min, k_max+1)]

    scores = []
    labels = []
    
    for n_clusters in range_n_clusters:    
        cluster = AgglomerativeClustering(n_clusters=n_clusters, affinity='euclidean', linkage='ward')  
        cluster.fit_predict(dataset)  
        cluster_labels = cluster.labels_
        labels.append(cluster_labels)
        write_to_csv(dataset, cluster_labels, 'clustering/hierarchical_with_' + str(n_clusters) + '_clusters.tsv')
              
    best = max_index(scores) + k_min
    
    for i in range(len(labels)):
        prefix = ''
        if (i + k_min == best):
            prefix = '_BEST'
        if silhouette == 'true':
            silihouette_draw(dataset, labels[i], i + k_min, 'clustering/silhouette_with_' + str(i + k_min) + prefix + '_clusters.png')
     
    for i in range(len(labels)):
        if (i + k_min == best):
            labels = labels[i]
            predict = [x+1 for x in labels]
            classe = (pd.DataFrame(list(zip(dataset.index, predict)))).astype(str)
            classe.to_csv(best_cluster, sep = '\t', index = False, header = ['Patient_ID', 'Class'])
            
    
############################# main ###########################################


def main():

    
    if args.cluster_type == 'dbscan':
        dbscan(X, args.eps, args.min_samples, args.best_cluster)
        
    if args.cluster_type == 'hierarchy':
        hierachical_agglomerative(X, args.k_min, args.k_max, args.best_cluster, args.silhouette)
        
##############################################################################

if __name__ == "__main__":
    main()