Transforming 6 digit integer to a date when bad values are involved

gbrown

I have a dataset with chunks of bad data in it. To get a MFG date I pulled the first 6 digits of the serial # to get a "date" in an text format.

if the data is correct it should read out the correct date, but in some cases the data will have 06B378 or other letters or texts that should not be in there. How would I go about excluding the bad data as well as converting the good 6 digit data into a date format?

Let me know your thoughts or how I go about this. I tried to alter columns to integers and inherit bad values as null but it seems to not be catching all bad values.

Thanks

DashboardDude

Hi @gbrown ,

So you're going to go have to do 2 steps:

1) Determine which cells are truly numeric

2) Convert the numeric ones to dates

Here are the formulas to do the steps:

1) CASE 

WHEN STR_REMOVE_DIGITS(`Date_Temp`)='' THEN 1 END

2) CASE 

WHEN `Numeric Format Flag`=1 

THEN CAST(CONCAT('20',RIGHT(`Date_Temp`,2),'-',left(`Date_Temp`,2),'-',SUBSTRING(`Date_Temp`,3,2)) as date)

END

Even better! Here's a video of me explaing what all this mumbo jumbo is: https://www.loom.com/share/ebf8d97b0ec94d458b6d09813dc93e2c

Hope that helps,

John Le

More video solutions at https://www.dashboarddudes.com/pantry

MichelleH

@gbrown Another option is to use the following formula in Magic ETL:

STR_TO_DATE(`Date_Temp`,'%m%d%y')