Using python this python notebook will automatically identify fraudulent credit card transactions utilising processes in machine learning.
As economic crime grows the concern for fraudulent activity has risen. As a result, the recent developments in deep learning, artificial intelligence and machine learning methods has facilitated many means for automating such a process. Through a systematic identification of outliers, developing this pythonic approach, potential anomalies can be understood and dealt with accordingly.
# Target variable visual comparison
fc = train_trn['isFraud'].value_counts(normalize = True).to_frame()
fc.plot.bar()
fc.T
# Correlation Heatmap of features C1 to C14
ccols = ['C%d' % number for number in range(1,15)]
plt.figure(figsize = (10,5))
corr = train_trn[['isFraud'] + ccols].corr()
sns.heatmap(corr, annot = True, fmt = '.2f')