Updated

classification.py

@@ -134,18 +134,27 @@ def validation(i):
 #X = X[(X['Trans'] > 2) & (X['Received'] > 200)]
 #X1 = X[X['Final bal'] <= 1]
 #X.drop(X1.index.values, axis=0, inplace=True)
-
+formula_address = [] 
+for i in range(len(X)):
+    rec = str(int(X.iloc[i]['Received']))
+    rec = len(rec)
+    trans = str(X.iloc[i]['Trans'])
+    trans = len(trans)
+    max_val = max([rec, trans])
+    min_val = min([rec, trans])
+    calc = min_val - (math.e/max_val)
+    formula_address.append(calc)
 X = X.reset_index()
 #X['Final bal'] = ((X['Final bal']) / (X['Trans']))
-X['Received'] = (X['Received'] - X['Trans']) / (X['Trans']+X['Received'])
-#X['Sent'] = (X['Sent']) / (X['Sent']+X['Received'])
+X['Received'] = abs(X['Received'] - X['Trans']) / (X['Trans']+X['Received'])
+X['ins'] = pd.Series(formula_address)
 #X['outs'] = ((X['ins'] - X['outs']) / (X['outs']+X['ins']))
 
 addr = list(X['Address'])
-X = X.drop(['Address', 'Trans', 'ins', 'index', 'outs', 'Sent', 'Final bal'], axis=1)
+#X = X.drop(['Address', 'Trans', 'index', 'outs', 'Sent', 'Final bal'], axis=1)
 
-clustering = KMeans(n_clusters = 2, random_state=11) # 3 for groups and 5 are random points
-clustering.fit(X)    
+clustering = KMeans(n_clusters = 3, random_state=7) # 3 for groups and 5 are random points
+clustering.fit(X[['Received', 'ins']])    
 
 pred_labels = clustering.labels_
 

classification_v2.py

@@ -0,0 +1,92 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Created on Wed Nov 21 12:45:20 2018
+
+@author: agile
+"""
+
+import blockchain
+from sklearn.cluster import KMeans
+from datetime import datetime
+import pandas as pd
+import math
+import pickle
+
+
+
+
+def validation(i):
+    address_details = pd.DataFrame(columns=['Address', 'Trans', 'Final bal', 'Received', 'Sent'])
+    for j in i:
+        try:
+            temp_addr = blockchain.blockexplorer.get_address(j)
+            n_tx = temp_addr.n_tx
+            final_bal = temp_addr.final_balance/100000000
+            recd = temp_addr.total_received/100000000
+            sent = temp_addr.total_sent/100000000
+
+            temp = {'Address': j, 'Trans': n_tx, 'Final bal': final_bal, 'Received': recd, 'Sent': sent}
+            address_details = address_details.append(temp , ignore_index=True)   
+        except Exception as e:
+            print(e, j)
+    return address_details 
+
+
+clustering = pickle.load(open("cluster.sav", 'rb'))
+X = validation(['37k5U5xQJkojFHkHVgrY7Dq7xtJJNSqVid', '17YyZSNFt31pzGXfZtrzs7Y5Nd56rG2uU5', 
+                '39vi392JLo4ksHUQrajddPJoK1wH8rmUyS', '1CSvNmJ4kCgQr9ubPaerrdhx7yUdLgKyKo',
+                '1JNW9w4kPiiBgaEW12xT5kiHxQgZxFRTQ4'])
+
+print(datetime.now())
+print('--------------------------------')
+
+intermediate_address = X[(X['Trans'] <= 5) & (X['Final bal'] <= 10)]
+intermediate_address.Sent = 'intermediate address'
+X.drop(intermediate_address.index.values, axis=0, inplace=True)
+
+still_waiting = X[X['Trans'] == 1]
+still_waiting.Sent = 'still waiting'
+X.drop(still_waiting.index.values, axis=0, inplace=True)
+cold_store = X[abs(X['Final bal'] - X['Received']) <= 10]
+cold_store.Sent = 'Cold Storage'
+X.drop(cold_store.index.values, axis=0, inplace=True)
+formula_address = [] 
+for i in range(len(X)):
+    rec = str(int(X.iloc[i]['Received']))
+    rec = len(rec)
+    trans = str(X.iloc[i]['Trans'])
+    trans = len(trans)
+    max_val = max([rec, trans])
+    min_val = min([rec, trans])
+    calc = min_val - (math.e/max_val)
+    formula_address.append(calc)
+X = X.reset_index()
+
+X['Received'] = abs(X['Received'] - X['Trans']) / (X['Trans']+X['Received'])
+X['Sent'] = pd.Series(formula_address)
+
+addr = list(X['Address'])
+predictions = clustering.predict(X[['Received', 'Sent']])
+
+final_labels=pd.DataFrame(columns=['Address', 'Label'])
+
+for i in range(len(addr)):
+    label = ''
+    if predictions[i] == 0:
+        label = 'Other'
+    elif predictions[i] == 1:
+        label = 'Light Exchange'
+    else:
+        label = 'Main Exhange'
+
+    final_labels = final_labels.append({'Address': addr[i], 'Label': label}, ignore_index=True)
+
+for i in intermediate_address, cold_store, still_waiting:
+    i = i[['Address', 'Sent']]
+    i.columns = ['Address', 'Label']
+    try:
+       final_labels = pd.concat([final_labels, i], ignore_index=True)
+    except:
+        1
+