The trick is to make sure your columns are recognized as dates, otherwise they will be evaluated as strings.

"to_date(<datecol>, <date_format>) " does exactly this.

datecol is your column containing the date values, and date_format is the format in which your existing date is formatted. In your case "MM/dd/yyyy".

This will return values of spark dateType.

When they are recognized as dates, you can start calculating with them.