find forward max flow correlation path

max_flow_correlation.py

def find_max_flow_pairs(corr_matrix):
    #corr_matrix = correlations_masked.copy()
    #def find_max_flow_pairs(corr_matrix):
    sorted_melted_correlations = corr_matrix.reset_index().melt(id_vars='index').replace('',np.nan).sort_values(by='value',ascending=False)
    #sorted_melted_correlations = corr_matrix.sort_values(by=['value'],ascending=False)
    sorted_melted_correlations_copy = sorted_melted_correlations.copy()
    max_flow_pairs = []
    nextmaxsorter = ''
    while (len(sorted_melted_correlations_copy)>1):

        if(nextmaxsorter==''):
            #print("if")
            max_score_row = sorted_melted_correlations_copy[sorted_melted_correlations_copy['value'] == sorted_melted_correlations_copy['value'].max()]
            paira = max_score_row['index'].values[0]
            pairb = max_score_row['Symbol'].values[0]
            score = max_score_row['value'].values[0]

            #print(max_score_row)
            #remove paira and pairb from next max calculation
            index_to_remove = sorted_melted_correlations_copy.index[(sorted_melted_correlations_copy['index'] == paira) & (sorted_melted_correlations_copy['Symbol'] == pairb)].tolist()
            sorted_melted_correlations_copy.drop(index=index_to_remove, inplace=True)
            index_to_remove_2 = sorted_melted_correlations_copy.index[(sorted_melted_correlations_copy['index'] == pairb) & (sorted_melted_correlations_copy['Symbol'] == paira)].tolist()        
            sorted_melted_correlations_copy.drop(index=index_to_remove_2, inplace=True)

            max_score_rows_a = sorted_melted_correlations_copy[sorted_melted_correlations_copy['index']==paira].head(1)
            #print(max_score_rows_a)
            max_score_rows_b = sorted_melted_correlations_copy[sorted_melted_correlations_copy['index']==pairb].head(1)
            #print(max_score_rows_b)

            #reduce sorted_melted_correlations_copy by removing elements with matching index and symbol names of the matching pair name that had the higher value
            if(max_score_rows_a['value'].values[0]>max_score_rows_b['value'].values[0]):
                removing = paira
                following = pairb   
                nextmaxsorter = following
            else:                           
                removing = pairb
                following = paira            
                nextmaxsorter = following
            #print("removing:", removing)
            #print("following:", following)

            index_to_remove3 = sorted_melted_correlations_copy.index[(sorted_melted_correlations_copy['index'] == removing) | (sorted_melted_correlations_copy['Symbol'] == removing)].tolist()
            #print("remove index:", index_to_remove3)
            sorted_melted_correlations_copy.drop(index=index_to_remove3, inplace=True)

            #print(np.sum(sorted_melted_correlations_copy['index']==removing)+np.sum(sorted_melted_correlations_copy['Symbol']==removing))

            set_ = [removing, following, score]

        else:
            #print(nextmaxsorter)
            max_score_row = sorted_melted_correlations_copy[sorted_melted_correlations_copy['index']==nextmaxsorter].sort_values(by='value',ascending=False).head(1)
            removing = nextmaxsorter
            #print("removing:", removing)
            following = max_score_row['Symbol'].values[0]
            #print("following:", following)
            score = max_score_row['value'].values[0]

            #print(max_score_row)
            #remove paira and pairb from next max calculation
            index_to_remove = sorted_melted_correlations_copy.index[(sorted_melted_correlations_copy['index'] == removing) & (sorted_melted_correlations_copy['Symbol'] == following)].tolist()
            sorted_melted_correlations_copy.drop(index=index_to_remove, inplace=True)
            index_to_remove_2 = sorted_melted_correlations_copy.index[(sorted_melted_correlations_copy['index'] == following) & (sorted_melted_correlations_copy['Symbol'] == removing)].tolist()
            sorted_melted_correlations_copy.drop(index=index_to_remove_2, inplace=True)
            #print('outside removing',removing)
            index_to_remove3 = sorted_melted_correlations_copy.index[(sorted_melted_correlations_copy['index'] == removing) | (sorted_melted_correlations_copy['Symbol'] == removing)].tolist()
            #print("remove index:", index_to_remove3)
            sorted_melted_correlations_copy.drop(index=index_to_remove3, inplace=True)
            #print(np.sum(sorted_melted_correlations_copy['index']==removing)+np.sum(sorted_melted_correlations_copy['Symbol']==removing))

            nextmaxsorter = following
            set_ = [removing, following, score]
            #print("following:", following)
            #print(max_score_row)
        max_flow_pairs.append([*set_,len(sorted_melted_correlations_copy)])
        #print(set_)
    #max_flow_pairs.append([sorted_melted_correlations_copy['index'].values[0],sorted_melted_correlations_copy['Symbol'].values[0],sorted_melted_correlations_copy['value'].values[0],1])
    return(max_flow_pairs)