Reducing the size of the data

The dataset that we are working with contains over 6 million rows of data. Most machine learning algorithms will take a large amount of time to work with a dataset of this size. In order to make our execution time quicker, we will reduce the size of the dataset to 20,000 rows. We can do this by using the following code:

#Storing the fraudulent data into a dataframe

df_fraud = df[df['isFraud'] == 1]

#Storing the non-fraudulent data into a dataframe 

df_nofraud = df[df['isFraud'] == 0]

#Storing 12,000 rows of non-fraudulent data

df_nofraud = df_nofraud.head(12000)

#Joining both datasets together 

df = pd.concat([df_fraud, df_nofraud], axis = 0)

In the preceding code, the fraudulent rows are stored in one dataframe. This dataframe contains a little over 8,000 rows. The 12,000 non-fraudulent rows are stored in another dataframe, and the two dataframes are joined together using the concat method from pandas.

This results in a dataframe with a little over 20,000 rows, over which we can now execute our algorithms relatively quickly.