nunosempere.github.io/maths-prog/MachineLearningDemystified/CleaningUpData.py

37 lines
1.1 KiB
Python
Raw Normal View History

2019-10-09 18:40:56 +00:00
directory = '/home/nuno/Documents/Jobs/IDInsight'
import pandas as pd
import numpy as np
## Install the dataframe
insuranceDataFrame = pd.read_csv(directory + '/insurance.csv')
## Some functions for cleaning up, inspired by R's ifelse function
def ifelse1(x, listOfChecks, yesLabel, noLabel):
if x in listOfChecks:
return (yesLabel)
else:
return (noLabel)
def ifelse2(x,listOfChecks, listOfLabels):
n = len(listOfChecks)
for i in range(n):
if x == listOfChecks[i]:
return (listOfLabels[i])
return None
insuranceDataFrame['sex_numeric'] =insuranceDataFrame['sex'].apply(lambda x: ifelse1(x, np.array(['male']),1,0))
insuranceDataFrame['smoker_numeric'] =insuranceDataFrame['smoker'].apply(lambda x: ifelse1(x, np.array(['yes']),1,0))
insuranceDataFrame['region_numeric'] =insuranceDataFrame['region'].apply(lambda x: ifelse2(x, np.unique(insuranceDataFrame['region']), np.array([0,1,2,3])))
insuranceDataFrame = insuranceDataFrame.drop(["sex", "smoker", "region"],axis=1)
## We save the database
insuranceDataFrame.to_csv(directory +'insurance_clean_continuous.csv', index=False)