df[columns] = df[columns].replace(',', '', regex=True).apply(pd.to_numeric) / 1000
df.to_csv(directory + 'last.csv', index=False, header=True)
print(len(df))

read= pd.read_csv('last.csv')

print(len(read))