使用Python脚本将'String Feature'数据框架转换为Azure ML中的Float

Question

I am trying to understand how to convert azure ml String Feature data type into float using python script. my data set is contain "HH:MM" data time format. It recognized as String Feature like the following img:

I want to convert it into float type which will divide the timestamp by 84600 ( 24 hour) so 17:30 will be converted into 0,729166666666667 , so I write python script to convert that. This is my script:

import pandas as pd
import numpy as np 

def timeToFloat(x):
    frt = [3600,60]
    data = str(x)
    result = float(sum([a*b for a,b in zip(frt, map(int,data.split(':')))]))/86400
    return result if isNotZero(x) else 0.0

def isNotZero(x):
    return (x is "0")

def azureml_main(dataframe1 = None):

    df = pd.DataFrame(dataframe1)
    df["Departure Time"] = pd.to_numeric(df["Departure Time"]).apply(timeToFloat)

    print(df["Departure Time"])

    return df,

When I run the script it was failed. Then I try to check whether it is str or not, but it returns None .

can we treat String Feature as String ? or how should I covert this data correctly?

Answer 1

The to_numeric conversion seems to be the problem, as there's no default parsing from string to number.

Does it work if you just use pd.apply(timeToFloat) ?

Roope - Microsoft Azure ML Team