Hi Sean, Even with MM for months it gives incorrect (but different this time) min value.
Le mar. 14 juin 2022 à 20:18, Sean Owen <sro...@gmail.com> a écrit : > Yes that is right. It has to be parsed as a date to correctly reason about > ordering. Otherwise you are finding the minimum string alphabetically. > > Small note, MM is month. mm is minute. You have to fix that for this to > work. These are Java format strings. > > On Tue, Jun 14, 2022, 12:32 PM marc nicole <mk1853...@gmail.com> wrote: > >> Hi, >> >> I want to identify a column of dates as such, the column has formatted >> strings in the likes of: "06-14-2022" (the format being mm-dd-yyyy) and get >> the minimum of those dates. >> >> I tried in Java as follows: >> >> if (dataset.filter(org.apache.spark.sql.functions.to_date( >>> dataset.col(colName), "mm-dd-yyyy").isNotNull()).select(colName).count() != >>> 0) { .... >> >> >> And to get the *min *of the column: >> >> Object colMin = >>> dataset.agg(org.apache.spark.sql.functions.min(org.apache.spark.sql.functions.to_date(dataset.col(colName), >>> "mm-dd-yyyy"))).first().get(0); >> >> // then I cast the *colMin *to string. >> >> To note that if i don't apply *to_date*() to the target column then the >> result will be erroneous (i think Spark will take the values as string and >> will get the min as if it was applied on an alphabetical string). >> >> Any better approach to accomplish this? >> Thanks. >> >