first_implementation_of_RNN

matching_chains_classificationRNNS.py

@@ -0,0 +1,602 @@
+import numpy as np
+from sklearn.metrics.pairwise import cosine_similarity
+import statistics
+import pandas as pd
+from sklearn.ensemble import RandomForestClassifier
+from sklearn.metrics import classification_report, confusion_matrix, \
+accuracy_score, roc_auc_score, roc_curve, f1_score, recall_score, precision_score
+import matplotlib.pyplot as plt
+import copy
+from sklearn import preprocessing, tree
+from sklearn.linear_model import LogisticRegression, LinearRegression
+from sklearn.tree import DecisionTreeClassifier
+from scipy.spatial import distance
+from sklearn.naive_bayes import GaussianNB
+import itertools
+import os
+from sklearn.model_selection import train_test_split
+import matplotlib.pyplot as plt
+import random
+from sklearn.utils import shuffle
+from imblearn.under_sampling import NearMiss,TomekLinks
+from imblearn.over_sampling import SMOTE
+from collections import Counter
+from imblearn.combine import SMOTETomek, SMOTEENN
+from sklearn.model_selection import StratifiedKFold
+from imblearn.pipeline import make_pipeline
+
+from matplotlib import pyplot
+from scipy import interp
+from sklearn.metrics import roc_curve,auc
+
+#keras
+from keras.models import Sequential
+from keras.layers import Dense, SimpleRNN, LSTM
+
+# Read ComE node embs per timestep [id, emb]
+
+folder = os.listdir('ComE_per_timestep/embs')
+path = 'ComE_per_timestep/embs'
+ComE_id_embs = []
+for file in folder:
+    ComE_id_embs.append(np.genfromtxt(os.path.join(path, file), dtype=None).tolist())
+
+# Read ComE labels per timestep
+
+folder = os.listdir('ComE_per_timestep/labels_pred')
+path = 'ComE_per_timestep/labels_pred'
+ComE_lbls = []
+for file in folder:
+    ComE_lbls.append(np.genfromtxt(os.path.join(path, file), dtype=None).tolist())
+
+# Node ids per timestep
+
+node_ids = []
+for step in ComE_id_embs:
+    tmp = [id_emb[0] for id_emb in step]
+    node_ids.append(tmp)
+
+# [Node_id, clr] per timestep
+
+id_clr = []
+for i in range(len(node_ids)):
+    tmp = {}
+    for ind,node in enumerate(node_ids[i]):
+        tmp[node] = ComE_lbls[i][ind]
+    id_clr.append(tmp)
+
+# Clustered nodes per timestep
+clustered_nodes_init = []
+for ind,i in enumerate(id_clr):
+    clrids_uniq = set(i.values())
+    d = {}
+    for clrid in clrids_uniq:
+        d[clrid] = [k for k in i.keys() if i[k] == clrid]
+    clustered_nodes_init.append(d)
+
+clustered_nodes = []
+for s in clustered_nodes_init:
+    per_step = []
+    for k,v in sorted(s.items()):
+        per_step.append(v)
+    clustered_nodes.append(per_step)
+                    
+# ------------------------------ READ FEATURES -------------------------------
+
+# ComE FEATURES
+
+folder = os.listdir('ComE_features_per_timestep/')
+path = 'ComE_features_per_timestep/'
+id_ComE_feats_clr = []
+id_ComE_feats_out = []
+id_ComE_feats_gbl = []
+id_ComE_feats_clrout = []
+id_ComE_feats_clrgbl = []
+id_ComE_feats_all = []
+for file in folder:
+    df_ComE = pd.read_csv(os.path.join(path,file), names=['node_id', \
+        'distin_med_eucl', 'distin_med_cos', 'distin_med_l1',\
+        'distout_med_eucl', 'distout_med_cos', 'distout_med_l1',\
+        'distin_eucl_max', 'distin_eucl_min', 'distin_eucl_avg',\
+        'distin_cos_max', 'distin_cos_min', 'distin_cos_avg',\
+        'distin_l1_max', 'distin_l1_min', 'distin_l1_avg',\
+        'distout_eucl_max', 'distout_eucl_min', 'distout_eucl_avg',\
+        'distout_cos_max', 'distout_cos_min', 'distout_cos_avg',\
+        'distout_l1_max', 'distout_l1_min', 'distout_l1_avg', \
+        'dist_glob_max_eucl', 'dist_glob_min_eucl', 'dist_glob_avg_eucl', \
+        'dist_glob_max_cos', 'dist_glob_min_cos', 'dist_glob_avg_cos', \
+        'dist_glob_max_l1', 'dist_glob_min_l1',  'dist_glob_avg_l1'], skiprows=1)
+    df_ComE_clr = df_ComE[['node_id', 'distin_med_eucl', \
+                          'distin_eucl_max', 'distin_eucl_min', 'distin_eucl_avg']]
+    df_ComE_out = df_ComE[['node_id', 'distout_med_eucl', \
+                        'distout_eucl_max', 'distout_eucl_min', 'distout_eucl_avg']]
+    df_ComE_gbl = df_ComE[['node_id', 'distout_med_eucl', \
+                        'dist_glob_max_eucl', 'dist_glob_min_eucl', 'dist_glob_avg_eucl']]
+    df_ComE_clrout = df_ComE[['node_id', 'distin_med_eucl', 'distout_med_eucl', \
+                          'distin_eucl_max', 'distin_eucl_min', 'distin_eucl_avg', \
+                        'distout_eucl_max', 'distout_eucl_min', 'distout_eucl_avg']]
+    df_ComE_clrgbl = df_ComE[['node_id', 'distin_med_eucl', \
+                          'distin_eucl_max', 'distin_eucl_min', 'distin_eucl_avg', \
+                        'dist_glob_max_eucl', 'dist_glob_min_eucl', 'dist_glob_avg_eucl']]
+    df_ComE_all = df_ComE[['node_id', 'distin_med_eucl', 'distout_med_eucl', \
+                          'distin_eucl_max', 'distin_eucl_min', 'distin_eucl_avg', \
+                        'distout_eucl_max', 'distout_eucl_min', 'distout_eucl_avg', \
+                        'dist_glob_max_eucl', 'dist_glob_min_eucl', 'dist_glob_avg_eucl']]
+    df_ComE_clr_lst = df_ComE_clr.values.tolist()
+    df_ComE_out_lst = df_ComE_out.values.tolist()
+    df_ComE_gbl_lst = df_ComE_gbl.values.tolist()
+    df_ComE_clrout_lst = df_ComE_clrout.values.tolist()
+    df_ComE_clrgbl_lst = df_ComE_clrgbl.values.tolist()
+    df_ComE_all_lst = df_ComE_all.values.tolist()
+    id_ComE_feats_clr.append(df_ComE_clr_lst)
+    id_ComE_feats_out.append(df_ComE_out_lst)
+    id_ComE_feats_gbl.append(df_ComE_gbl_lst)
+    id_ComE_feats_clrout.append(df_ComE_clrout_lst)
+    id_ComE_feats_clrgbl.append(df_ComE_clrgbl_lst)
+    id_ComE_feats_all.append(df_ComE_all_lst)
+#sort by node id
+for i in id_ComE_feats_clr:
+    i.sort()
+for i in id_ComE_feats_out:
+    i.sort()
+for i in id_ComE_feats_gbl:
+    i.sort()
+for i in id_ComE_feats_clrout:
+    i.sort()
+for i in id_ComE_feats_clrgbl:
+    i.sort()
+for i in id_ComE_feats_all:
+    i.sort()
+
+# Classic FEATURES
+
+folder = os.listdir('classic_features_per_timestep/classic_features')
+path = 'classic_features_per_timestep/classic_features'
+id_classic_clr = []
+id_classic_gbl = []
+id_classic_all = []
+id_classic_nodeg = []
+for file in folder:
+    df_classic = pd.read_csv(os.path.join(path,file), names=['node_id', \
+    'degree', 'betweenness', 'closeness', 'eigenvector', \
+    'degree_ntwk', 'betweenness_ntwk', 'closeness_ntwk', 'eigenvector_ntwk'], \
+    skiprows=1)
+    df_classic_clr = df_classic[['node_id', \
+                    'degree', 'betweenness', 'closeness', 'eigenvector']]
+    df_classic_gbl = df_classic[['node_id', \
+    'degree_ntwk', 'betweenness_ntwk', 'closeness_ntwk', 'eigenvector_ntwk']]
+    df_classic_nodeg = pd.read_csv(os.path.join(path,file), names=['node_id', \
+    'betweenness', 'closeness', 'eigenvector', \
+    'betweenness_ntwk', 'closeness_ntwk', 'eigenvector_ntwk'], \
+    skiprows=1)
+    df_classic_all_lst = df_classic.values.tolist()
+    df_classic_clr_lst = df_classic_clr.values.tolist()
+    id_classic_gbl_lst = df_classic_gbl.values.tolist()
+    id_classic_nodeg_lst = df_classic_nodeg.values.tolist()
+    id_classic_all.append(df_classic_all_lst)
+    id_classic_clr.append(df_classic_clr_lst)
+    id_classic_gbl.append(id_classic_gbl_lst)
+    id_classic_nodeg.append(id_classic_nodeg_lst)
+#sort by node id
+for i in id_classic_all:
+    i.sort()
+for i in id_classic_clr:
+    i.sort()
+for i in id_classic_gbl:
+    i.sort()
+for i in id_classic_nodeg:
+    i.sort()
+
+id_combo_ComE_clrout_classic_all = []
+for ind,s in enumerate(id_ComE_feats_clrout):
+    temp = []
+    for inx,row in enumerate(s):
+        tmp = row[:]
+        tmp.extend(id_classic_all[ind][inx][1:])
+        temp.append(tmp)
+    id_combo_ComE_clrout_classic_all.append(temp)
+
+#-------------------------------- MATCHING ------------------------------------
+
+# [clr_x_tn, clr_y_tn+1, common_nodes_tn_tn+1]
+#print(clustered_nodes[0])
+matching = []
+a = 0
+while a<len(clustered_nodes)-1:
+    matching_two = []
+    for indcurr,clrcurr in enumerate(clustered_nodes[a]):
+        tmp = []
+        for indnxt,clrnxt in enumerate(clustered_nodes[a+1]):
+            num_of_common = len(list(set(clrcurr)&set(clrnxt)))
+            tmp.append([indcurr,indnxt,num_of_common])
+        tmp_max = max(item[-1] for item in tmp)
+        for t in tmp:
+            if t[-1] == tmp_max:
+                maxtmp = t
+        matching_two.append(maxtmp)
+    matching.append(matching_two)
+    a += 1
+#print(matching,a)
+
+
+
+
+#--------------------------------- CHAINS -------------------------------------
+#************ SCD    #Stay-Change-Drop
+# 2-chain
+def twoChain_scd(features):
+    two_chain_scd = []
+    for ind,step in enumerate(matching[:-1]):
+        per_step = []
+        for inx,clr in enumerate(step):
+            for nodeid in clustered_nodes[ind][clr[0]]:
+                tmp = [nodeid]
+                for idfeatures in features[ind]:
+                    if nodeid == idfeatures[0]:
+                        tmp.extend(idfeatures[1:])
+                if nodeid in clustered_nodes[ind+1][clr[1]]:
+                    tmp.append(0)#stay
+                    #for idfeatures in features[ind+1]:  remove second set of features
+                        #if nodeid == idfeatures[0]:
+                            #tmp.extend(idfeatures[1:])
+                    '''for cl in matching[ind+1]:   
+                        if nodeid in clustered_nodes[ind+1][cl[0]]:
+                            if nodeid in clustered_nodes[ind+2][cl[1]]:
+                                tmp.append(0)#stay
+                                break
+                            elif nodeid in node_ids[ind+2]:
+                                tmp.append(1)#move
+                                break
+                            else:
+                                tmp.append(2)#drop
+                                break'''
+                elif nodeid in node_ids[ind+1]:
+                    tmp.append(1)#move
+                    #for idfeatures in features[ind+1]:
+                        #if nodeid == idfeatures[0]:
+                            #tmp.extend(idfeatures[1:])
+                    #for cl in matching[ind+1]:
+                        #if nodeid in clustered_nodes[ind+1][cl[0]]:
+                            #if nodeid in clustered_nodes[ind+2][cl[1]]:
+                                #tmp.append(0)#stay
+                               # break
+                            #elif nodeid in node_ids[ind+2]:
+                               # tmp.append(1)#move
+                               # break
+                           # else:
+                              #  tmp.append(2)#drop
+                               # break
+                else:
+                  #  tmp.extend([-1]*(len(features[0][0][1:])+1))  remove extend vasia
+                    tmp.append(2)#drop
+                per_step.append(tmp)
+        two_chain_scd.append(per_step)
+    return(two_chain_scd)
+
+def chains_scd(prev_chain_scd, features, a):
+    curr_chain_scd = copy.deepcopy(prev_chain_scd[:-1])
+    for ind,step in enumerate(curr_chain_scd):
+        for row in step:
+            if row[-1] == 0 or row[-1] == 1:
+                for idfeatures in features[ind+2+a]:
+                    if row[0] == idfeatures[0]:
+                        row.extend(idfeatures[1:])
+                for cl in matching[ind+2+a]:
+                    if row[0] in clustered_nodes[ind+2+a][cl[0]]:
+                        if row[0] in clustered_nodes[ind+3+a][cl[1]]:
+                            row.append(0)#stay
+                            break
+                        elif row[0] in node_ids[ind+3+a]:
+                            row.append(1)#move
+                            break
+                        else:
+                            row.append(2)#drop
+                            break
+            else:
+                row[-1:-1] = [-1]*(len(features[0][0][1:])+1)#add -1*(#feats + ev)
+    return(curr_chain_scd)
+
+# ----------------------------------------------------------------------------
+
+#************ SL
+def chains_sl(chainsSCD):         #Stay-Leave
+    chainsSL = copy.deepcopy(chainsSCD)
+    for row in chainsSL:
+        if row[-1] == 2:
+            row[-1] = 1
+    return(chainsSL)
+
+# ----------------------------------------------------------------------------
+
+#************ SC        #Stay-Change
+def chains_sc(chainsSCD):
+    chainsSC = []
+    for row in chainsSCD:
+        if row[-1] != 2:
+            chainsSC.append(row)
+    return(chainsSC)
+
+# ----------------------------------------------------------------------------
+
+def per_chain_all_chains_scd(feats):
+    two_chain_scd = twoChain_scd(feats)
+    three_chain_scd = chains_scd(two_chain_scd, feats, 0)
+    four_chain_scd = chains_scd(three_chain_scd, feats, 1)
+    five_chain_scd = chains_scd(four_chain_scd, feats, 2)
+    six_chain_scd = chains_scd(five_chain_scd, feats, 3)
+    seven_chain_scd = chains_scd(six_chain_scd, feats, 4)
+    eight_chain_scd = chains_scd(seven_chain_scd, feats, 5)
+    nine_chain_scd = chains_scd(eight_chain_scd, feats, 6)
+    two_chain_scd = [row for s in two_chain_scd for row in s]#flat
+    three_chain_scd = [row for s in three_chain_scd for row in s]#flat
+    four_chain_scd = [row for s in four_chain_scd for row in s]#flat
+    five_chain_scd = [row for s in five_chain_scd for row in s]#flat
+    six_chain_scd = [row for s in six_chain_scd for row in s]#flat
+    seven_chain_scd = [row for s in seven_chain_scd for row in s]#flat
+    eight_chain_scd = [row for s in eight_chain_scd for row in s]#flat
+    nine_chain_scd = [row for s in nine_chain_scd for row in s]#flat
+    # merge chains
+    all_chains_scd = []
+    all_chains_scd.append(two_chain_scd)
+    all_chains_scd.append(three_chain_scd)
+    all_chains_scd.append(four_chain_scd)
+    all_chains_scd.append(five_chain_scd)
+    all_chains_scd.append(six_chain_scd)
+    all_chains_scd.append(seven_chain_scd)
+    all_chains_scd.append(eight_chain_scd)
+    all_chains_scd.append(nine_chain_scd)
+    all_chains_scd = [row for chain in all_chains_scd for row in chain]
+    return(two_chain_scd, three_chain_scd, four_chain_scd, five_chain_scd, \
+           six_chain_scd, seven_chain_scd, eight_chain_scd, nine_chain_scd, \
+               all_chains_scd)
+
+# ----------------------------------------------------------------------------
+
+# CHAINS ----------------------
+# ComE
+
+# clr
+two_chain_ComE_clr_scd, three_chain_ComE_clr_scd, four_chain_ComE_clr_scd, \
+    five_chain_ComE_clr_scd, six_chain_ComE_clr_scd, seven_chain_ComE_clr_scd, \
+    eight_chain_ComE_clr_scd, nine_chain_ComE_clr_scd, \
+    chains_ComE_clr_scd = per_chain_all_chains_scd(id_ComE_feats_clr)
+# per chain
+two_chain_ComE_clr_sl = chains_sl(two_chain_ComE_clr_scd)
+two_chain_ComE_clr_sc = chains_sc(two_chain_ComE_clr_scd)
+three_chain_ComE_clr_sl = chains_sl(three_chain_ComE_clr_scd)
+three_chain_ComE_clr_sc = chains_sc(three_chain_ComE_clr_scd)
+four_chain_ComE_clr_sl = chains_sl(four_chain_ComE_clr_scd)
+four_chain_ComE_clr_sc = chains_sc(four_chain_ComE_clr_scd)
+five_chain_ComE_clr_sl = chains_sl(five_chain_ComE_clr_scd)
+five_chain_ComE_clr_sc = chains_sc(five_chain_ComE_clr_scd)
+six_chain_ComE_clr_sl = chains_sl(six_chain_ComE_clr_scd)
+six_chain_ComE_clr_sc = chains_sc(six_chain_ComE_clr_scd)
+seven_chain_ComE_clr_sl = chains_sl(seven_chain_ComE_clr_scd)
+seven_chain_ComE_clr_sc = chains_sc(seven_chain_ComE_clr_scd)
+eight_chain_ComE_clr_sl = chains_sl(eight_chain_ComE_clr_scd)
+eight_chain_ComE_clr_sc = chains_sc(eight_chain_ComE_clr_scd)
+nine_chain_ComE_clr_sl = chains_sl(nine_chain_ComE_clr_scd)
+nine_chain_ComE_clr_sc = chains_sc(nine_chain_ComE_clr_scd)
+# SL
+chains_ComE_clr_sl = chains_sl(chains_ComE_clr_scd)
+# SC
+chains_ComE_clr_sc = chains_sc(chains_ComE_clr_scd)
+
+# out
+two_chain_ComE_out_scd, three_chain_ComE_out_scd, four_chain_ComE_out_scd, \
+    five_chain_ComE_out_scd, six_chain_ComE_out_scd, seven_chain_ComE_out_scd, \
+    eight_chain_ComE_out_scd, nine_chain_ComE_out_scd, \
+    chains_ComE_out_scd = per_chain_all_chains_scd(id_ComE_feats_out)
+# per chain
+two_chain_ComE_out_sl = chains_sl(two_chain_ComE_out_scd)
+two_chain_ComE_out_sc = chains_sc(two_chain_ComE_out_scd)
+three_chain_ComE_out_sl = chains_sl(three_chain_ComE_out_scd)
+three_chain_ComE_out_sc = chains_sc(three_chain_ComE_out_scd)
+four_chain_ComE_out_sl = chains_sl(four_chain_ComE_out_scd)
+four_chain_ComE_out_sc = chains_sc(four_chain_ComE_out_scd)
+five_chain_ComE_out_sl = chains_sl(five_chain_ComE_out_scd)
+five_chain_ComE_out_sc = chains_sc(five_chain_ComE_out_scd)
+six_chain_ComE_out_sl = chains_sl(six_chain_ComE_out_scd)
+six_chain_ComE_out_sc = chains_sc(six_chain_ComE_out_scd)
+seven_chain_ComE_out_sl = chains_sl(seven_chain_ComE_out_scd)
+seven_chain_ComE_out_sc = chains_sc(seven_chain_ComE_out_scd)
+eight_chain_ComE_out_sl = chains_sl(eight_chain_ComE_out_scd)
+eight_chain_ComE_out_sc = chains_sc(eight_chain_ComE_out_scd)
+nine_chain_ComE_out_sl = chains_sl(nine_chain_ComE_out_scd)
+nine_chain_ComE_out_sc = chains_sc(nine_chain_ComE_out_scd)
+# SL
+chains_ComE_out_sl = chains_sl(chains_ComE_out_scd)
+# SC
+chains_ComE_out_sc = chains_sc(chains_ComE_out_scd)
+
+# clrout
+two_chain_ComE_clrout_scd, three_chain_ComE_clrout_scd, four_chain_ComE_clrout_scd, \
+    five_chain_ComE_clrout_scd, six_chain_ComE_clrout_scd, seven_chain_ComE_clrout_scd, \
+    eight_chain_ComE_clrout_scd, nine_chain_ComE_clrout_scd, \
+    chains_ComE_clrout_scd = per_chain_all_chains_scd(id_ComE_feats_clrout)
+# per chain
+two_chain_ComE_clrout_sl = chains_sl(two_chain_ComE_clrout_scd)
+two_chain_ComE_clrout_sc = chains_sc(two_chain_ComE_clrout_scd)
+three_chain_ComE_clrout_sl = chains_sl(three_chain_ComE_clrout_scd)
+three_chain_ComE_clrout_sc = chains_sc(three_chain_ComE_clrout_scd)
+four_chain_ComE_clrout_sl = chains_sl(four_chain_ComE_clrout_scd)
+four_chain_ComE_clrout_sc = chains_sc(four_chain_ComE_clrout_scd)
+five_chain_ComE_clrout_sl = chains_sl(five_chain_ComE_clrout_scd)
+five_chain_ComE_clrout_sc = chains_sc(five_chain_ComE_clrout_scd)
+six_chain_ComE_clrout_sl = chains_sl(six_chain_ComE_clrout_scd)
+six_chain_ComE_clrout_sc = chains_sc(six_chain_ComE_clrout_scd)
+seven_chain_ComE_clrout_sl = chains_sl(seven_chain_ComE_clrout_scd)
+seven_chain_ComE_clrout_sc = chains_sc(seven_chain_ComE_clrout_scd)
+eight_chain_ComE_clrout_sl = chains_sl(eight_chain_ComE_clrout_scd)
+eight_chain_ComE_clrout_sc = chains_sc(eight_chain_ComE_clrout_scd)
+nine_chain_ComE_clrout_sl = chains_sl(nine_chain_ComE_clrout_scd)
+nine_chain_ComE_clrout_sc = chains_sc(nine_chain_ComE_clrout_scd)
+# SL
+chains_ComE_clrout_sl = chains_sl(chains_ComE_clrout_scd)
+# SC
+chains_ComE_clrout_sc = chains_sc(chains_ComE_clrout_scd)
+
+
+
+# ----------------------------------------------------------------------------
+
+# Classic
+#clr
+# SCD
+two_chain_classic_clr_scd, three_chain_classic_clr_scd, four_chain_classic_clr_scd, \
+    five_chain_classic_clr_scd, six_chain_classic_clr_scd, seven_chain_classic_clr_scd, \
+        eight_chain_classic_clr_scd, nine_chain_classic_clr_scd, \
+    chains_classic_clr_scd = per_chain_all_chains_scd(id_classic_clr)
+# per chain
+two_chain_classic_clr_sl = chains_sl(two_chain_classic_clr_scd)
+two_chain_classic_clr_sc = chains_sc(two_chain_classic_clr_scd)
+three_chain_classic_clr_sl = chains_sl(three_chain_classic_clr_scd)
+three_chain_classic_clr_sc = chains_sc(three_chain_classic_clr_scd)
+four_chain_classic_clr_sl = chains_sl(four_chain_classic_clr_scd)
+four_chain_classic_clr_sc = chains_sc(four_chain_classic_clr_scd)
+five_chain_classic_clr_sl = chains_sl(five_chain_classic_clr_scd)
+five_chain_classic_clr_sc = chains_sc(five_chain_classic_clr_scd)
+six_chain_classic_clr_sl = chains_sl(six_chain_classic_clr_scd)
+six_chain_classic_clr_sc = chains_sc(six_chain_classic_clr_scd)
+seven_chain_classic_clr_sl = chains_sl(seven_chain_classic_clr_scd)
+seven_chain_classic_clr_sc = chains_sc(seven_chain_classic_clr_scd)
+eight_chain_classic_clr_sl = chains_sl(eight_chain_classic_clr_scd)
+eight_chain_classic_clr_sc = chains_sc(eight_chain_classic_clr_scd)
+nine_chain_classic_clr_sl = chains_sl(nine_chain_classic_clr_scd)
+nine_chain_classic_clr_sc = chains_sc(nine_chain_classic_clr_scd)
+
+# SL
+chains_classic_clr_sl = chains_sl(chains_classic_clr_scd)
+# SC
+chains_classic_clr_sc = chains_sc(chains_classic_clr_scd)
+
+#gbl
+# SCD
+two_chain_classic_gbl_scd, three_chain_classic_gbl_scd, four_chain_classic_gbl_scd, \
+    five_chain_classic_gbl_scd, six_chain_classic_gbl_scd, seven_chain_classic_gbl_scd, \
+        eight_chain_classic_gbl_scd, nine_chain_classic_gbl_scd, \
+    chains_classic_gbl_scd = per_chain_all_chains_scd(id_classic_gbl)
+# per chain
+two_chain_classic_gbl_sl = chains_sl(two_chain_classic_gbl_scd)
+two_chain_classic_gbl_sc = chains_sc(two_chain_classic_gbl_scd)
+three_chain_classic_gbl_sl = chains_sl(three_chain_classic_gbl_scd)
+three_chain_classic_gbl_sc = chains_sc(three_chain_classic_gbl_scd)
+four_chain_classic_gbl_sl = chains_sl(four_chain_classic_gbl_scd)
+four_chain_classic_gbl_sc = chains_sc(four_chain_classic_gbl_scd)
+five_chain_classic_gbl_sl = chains_sl(five_chain_classic_gbl_scd)
+five_chain_classic_gbl_sc = chains_sc(five_chain_classic_gbl_scd)
+six_chain_classic_gbl_sl = chains_sl(six_chain_classic_gbl_scd)
+six_chain_classic_gbl_sc = chains_sc(six_chain_classic_gbl_scd)
+seven_chain_classic_gbl_sl = chains_sl(seven_chain_classic_gbl_scd)
+seven_chain_classic_gbl_sc = chains_sc(seven_chain_classic_gbl_scd)
+eight_chain_classic_gbl_sl = chains_sl(eight_chain_classic_gbl_scd)
+eight_chain_classic_gbl_sc = chains_sc(eight_chain_classic_gbl_scd)
+nine_chain_classic_gbl_sl = chains_sl(nine_chain_classic_gbl_scd)
+nine_chain_classic_gbl_sc = chains_sc(nine_chain_classic_gbl_scd)
+
+# SL
+chains_classic_gbl_sl = chains_sl(chains_classic_gbl_scd)
+# SC
+chains_classic_gbl_sc = chains_sc(chains_classic_gbl_scd)
+
+#all
+# SCD
+two_chain_classic_all_scd, three_chain_classic_all_scd, four_chain_classic_all_scd, \
+    five_chain_classic_all_scd, six_chain_classic_all_scd, seven_chain_classic_all_scd, \
+        eight_chain_classic_all_scd, nine_chain_classic_all_scd, \
+    chains_classic_all_scd = per_chain_all_chains_scd(id_classic_all)
+# per chain
+two_chain_classic_all_sl = chains_sl(two_chain_classic_all_scd)
+two_chain_classic_all_sc = chains_sc(two_chain_classic_all_scd)
+three_chain_classic_all_sl = chains_sl(three_chain_classic_all_scd)
+three_chain_classic_all_sc = chains_sc(three_chain_classic_all_scd)
+four_chain_classic_all_sl = chains_sl(four_chain_classic_all_scd)
+four_chain_classic_all_sc = chains_sc(four_chain_classic_all_scd)
+five_chain_classic_all_sl = chains_sl(five_chain_classic_all_scd)
+five_chain_classic_all_sc = chains_sc(five_chain_classic_all_scd)
+six_chain_classic_all_sl = chains_sl(six_chain_classic_all_scd)
+six_chain_classic_all_sc = chains_sc(six_chain_classic_all_scd)
+seven_chain_classic_all_sl = chains_sl(seven_chain_classic_all_scd)
+seven_chain_classic_all_sc = chains_sc(seven_chain_classic_all_scd)
+eight_chain_classic_all_sl = chains_sl(eight_chain_classic_all_scd)
+eight_chain_classic_all_sc = chains_sc(eight_chain_classic_all_scd)
+nine_chain_classic_all_sl = chains_sl(nine_chain_classic_all_scd)
+nine_chain_classic_all_sc = chains_sc(nine_chain_classic_all_scd)
+
+# SL
+chains_classic_all_sl = chains_sl(chains_classic_all_scd)
+# SC
+chains_classic_all_sc = chains_sc(chains_classic_all_scd)
+
+
+#RNN model
+def create_RNN(hidden_units, dense_units, input_shape, activation):
+    model=Sequential()
+    model.add(LSTM(hidden_units,input_shape=input_shape))
+    model.add(Dense(units=dense_units,activation=activation))
+    model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])
+    model.summary() 
+    return model
+
+# stratified kfold 
+def Classification(chain):
+    #print("chain1 = ",len(chain),chain[0])
+   
+    chain = shuffle(np.array(chain))
+    print("chain2 = ",chain.shape)
+    X = [i[1:-1] for i in chain.tolist()]
+    Y = [i[-1] for i in chain.tolist()]
+
+    # padding
+    longest = len(max(X,key=len))
+    print(longest)
+
+    for row in X:
+        while len(row)<longest:
+            row.append(-99)
+    #######
+    X = np.array(X)
+    Y = np.array(Y)
+    print('Y_dataset:', Counter(Y))
+    skf = StratifiedKFold(n_splits=5)
+    fold = 0
+    k=0
+    cvscores = []
+    for train_index, test_index in skf.split(X, Y):
+        X_train, X_test = X[train_index], X[test_index]
+        
+        y_train, y_test = Y[train_index], Y[test_index]
+        y_train_cnt = pd.DataFrame([Counter(y_train)]).transpose()
+        print(y_train_cnt)
+        y_train_cnt.sort_index(inplace=True)
+        print('y_train:', y_train_cnt)
+       # print("X",X_train.shape,X_train[0])
+
+        X_train_3d = X_train.reshape((X_train.shape[0], 4, 1))
+        X_test_3d = X_test.reshape((X_test.shape[0], 4, 1))
+       # print(X_test_3d)
+   
+        rnn_model = create_RNN(100,2,input_shape=(4,1),activation='sigmoid')
+
+        rnn_model.fit(X_train_3d, y_train, epochs=50, batch_size=5,verbose=2)
+        #test_pred = rnn_model.predict(X_test_3d)
+       
+        scores = rnn_model.evaluate(X_test_3d, y_test, verbose=0)
+        print("score",scores)
+        exit()
+        print("%s: %.2f%%" % (rnn_model.metrics_names[1], scores[1]*100))
+        cvscores.append(scores[1] * 100)
+    print("%.2f%% (+/- %.2f%%)" % (np.mean(cvscores), np.std(cvscores)))
+
+
+
+
+    # Classic features
+    # clr - 1 chain
+with open('results_details.csv','a') as fd:
+    fd.write('two_chain_classic_clr_sl'+'\n')
+with open('results.csv','a') as fd:
+    fd.write('two_chain_classic_clr_sl'+'\n')
+Classification(two_chain_classic_clr_sl)
+
+