In [1]:
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np
In [110]:
df = pd.read_csv('./Self_Organizing_Maps/Credit_Card_Applications.csv')
df.shape
Out[110]:
In [111]:
df.tail()
Out[111]:
In [112]:
(df['CustomerID'].unique()).shape
Out[112]:
In [113]:
sp = lambda *x: [i.shape for i in x]
In [114]:
X = df.iloc[:,:-1].values
y = df.iloc[:,-1].values
sp(X, y)
Out[114]:
In [115]:
X
Out[115]:
feature scaling¶
In [116]:
from sklearn.preprocessing import MinMaxScaler
sc = MinMaxScaler()
sc
Out[116]:
In [117]:
X = sc.fit_transform(X)
X.shape
Out[117]:
In [26]:
import sys
sys.path.insert(0,'./Self_Organizing_Maps/')
In [32]:
from minisom import MiniSom
som = MiniSom(10, 10, 15)
In [56]:
som.random_weights_init(X)
som.train_random(data=X, num_iteration=10000)
In [57]:
from pylab import bone, pcolor, colorbar, plot, show
%matplotlib inline
In [60]:
som.distance_map()
Out[60]:
In [79]:
plt.figure(figsize=((6,6)))
plt.imshow(som.distance_map())
plt.colorbar()
Out[79]:
In [72]:
import seaborn as sns
sns.heatmap(som.distance_map())
Out[72]:
In [89]:
plt.figure(figsize=((12,8)))
bone()
pcolor(som.distance_map().T)
colorbar()
markers = ['o','s']
colors=['r','g']
for i, x in enumerate(X):
w = som.winner(x)
plot(w[0]+.5,
w[1]+.5,
markers[y[i]],
markeredgecolor=colors[y[i]],
markerfacecolor='None',
markersize=10,
markeredgewidth=2)
show()
In [92]:
mappings=som.win_map(X)
len(mappings), type(mappings)
Out[92]:
In [94]:
# mappings
In [101]:
frauds = np.concatenate([mappings[(1,1)], mappings[(1,8)]])
frauds
Out[101]:
In [107]:
sc.inverse_transform?
In [108]:
X
Out[108]:
In [119]:
sc.inverse_transform(frauds)
Out[119]: