Data wrangling: strings to numbers

Many datasets have numerical values encoded in strings which need to be converted intonumbers for analysis

data = '40k'

data.split('k')

['40', '']

int(data.split('k')[0])*1000

40000

datalist = ['40k', '31k', '12k']

For use with Table columns or other array data¶

from datascience import *
import numpy as np

from datascience import *
t = Table().with_columns('index',[0,1,2],'amount',datalist)
t

t = t.with_columns('value',[int(data.split('k')[0])*1000 for data in datalist])
t

Float data embeded in string within table¶

datalist = ['4.01k', '3.11k', '1.25k']

[float(data.split('k')[0])*1000 for data in datalist]

[4010.0, 3110.0, 1250.0]

salary = Table().with_columns('position',['Data scientist','Chemist','Chemist','Biologist','Physicist','Finance'],
                              'salary',['75k','102k','99k','103k','99k','34k'])
salary

amount = [int(data.replace(',','').split('k')[0])*1000 for data in salary.column('salary')]
amount

[75000, 102000, 99000, 103000, 99000, 34000]

salary = salary.with_columns('salary',amount)
salary

Working with time strings¶

time_values = ['12:03:56', '01:04:23', '03:35:00']

t = t.with_columns('time',time_values,'hour',[int(data.split(':')[0]) for data in time_values])
t