If I create a table and specify a CSVSerde then all fields are being converted to string type.
hive> create table foo(a int, b double, c string) row format serde 'com.bizo.hive.serde.csv.CSVSerde' stored as textfile;
OK
Time taken: 0.22 seconds
hive> describe foo;
OK
a string from deserializer
b string from deserializer
c string from deserializer
Time taken: 0.063 seconds, Fetched: 3 row(s)
That Serde is from https://github.com/ogrodnek/csv-serde
If I try the serde 'org.apache.hadoop.hive.serde2.OpenCSVSerde' from this page https://cwiki.apache.org/confluence/display/Hive/CSV+Serde I saw the same thing. All fields are being changed to type string.
Hive version 1.2.1 Hadoop version 2.7.0 java version "1.7.0_80"