Spark Dataframe IN-ISIN-NOT IN

IN or NOT IN conditions are used in FILTER/WHERE or even in JOINS when we have to specify multiple possible values for any column. If the value is one of the values mentioned inside “IN” clause then it will qualify. It is opposite for “NOT IN” where the value must not be among any one present inside NOT IN clause.
So let’s look at the example for IN condition

scala> df_pres.filter($"pres_bs" in ("New York","Ohio","Texas")).select($"pres_name",$"pres_dob",$"pres_bs").show()
+--------------------+----------+--------+
|           pres_name|  pres_dob| pres_bs|
+--------------------+----------+--------+
|    Martin Van Buren|1782-12-05|New York|
|    Millard Fillmore|1800-01-07|New York|
|    Ulysses S. Grant|1822-04-27|    Ohio|
| Rutherford B. Hayes|1822-10-04|    Ohio|
|   James A. Garfield|1831-11-19|    Ohio|
|   Benjamin Harrison|1833-08-20|    Ohio|
|    William McKinley|1843-01-29|    Ohio|
|  Theodore Roosevelt|1858-10-27|New York|
| William Howard Taft|1857-09-15|    Ohio|
|   Warren G. Harding|1865-11-02|    Ohio|
|Franklin D. Roose...|1882-01-30|New York|
|Dwight D. Eisenhower|1890-10-14|   Texas|
|   Lyndon B. Johnson|1908-08-27|   Texas|
|        Donald Trump|1946-06-14|New York|
+--------------------+----------+--------+

Note: “in” method is not available in Spark 2.0. So prefer method is “isin”

Other way of writing it could be and the one which I prefer is by using isin function.

scala> df_pres.filter($"pres_bs".isin("New York","Ohio","Texas")).select($"pres_name",$"pres_dob",$"pres_bs").show()
+--------------------+----------+--------+
|           pres_name|  pres_dob| pres_bs|
+--------------------+----------+--------+
|    Martin Van Buren|1782-12-05|New York|
|    Millard Fillmore|1800-01-07|New York|
|    Ulysses S. Grant|1822-04-27|    Ohio|
| Rutherford B. Hayes|1822-10-04|    Ohio|
|   James A. Garfield|1831-11-19|    Ohio|
|   Benjamin Harrison|1833-08-20|    Ohio|
|    William McKinley|1843-01-29|    Ohio|
|  Theodore Roosevelt|1858-10-27|New York|
| William Howard Taft|1857-09-15|    Ohio|
|   Warren G. Harding|1865-11-02|    Ohio|
|Franklin D. Roose...|1882-01-30|New York|
|Dwight D. Eisenhower|1890-10-14|   Texas|
|   Lyndon B. Johnson|1908-08-27|   Texas|
|        Donald Trump|1946-06-14|New York|
+--------------------+----------+--------+

To use the condition as “NOT IN”, you can use negation (!) before the column name in the previous isin query.

scala> df_pres.filter(!$"pres_bs".isin("New York","Ohio","Texas")).select($"pres_name",$"pres_dob",$"pres_bs").show()
+--------------------+----------+--------------------+
|           pres_name|  pres_dob|             pres_bs|
+--------------------+----------+--------------------+
|   George Washington|1732-02-22|            Virginia|
|          John Adams|1735-10-30|       Massachusetts|
|    Thomas Jefferson|1743-04-13|            Virginia|
|       James Madison|1751-03-16|            Virginia|
|        James Monroe|1758-04-28|            Virginia|
|   John Quincy Adams|1767-07-11|       Massachusetts|
|      Andrew Jackson|1767-03-15|South/North Carolina|
|William Henry Har...|1773-02-09|            Virginia|
|          John Tyler|1790-03-29|            Virginia|
|       James K. Polk|1795-11-02|      North Carolina|
|      Zachary Taylor|1784-11-24|            Virginia|
|     Franklin Pierce|1804-11-23|       New Hampshire|
|      James Buchanan|1791-04-23|        Pennsylvania|
|     Abraham Lincoln|1809-02-12|            Kentucky|
|      Andrew Johnson|1808-12-29|      North Carolina|
|   Chester A. Arthur|1829-10-05|             Vermont|
|    Grover Cleveland|1837-03-18|          New Jersey|
|    Grover Cleveland|1837-03-18|          New Jersey|
|      Woodrow Wilson|1856-12-28|            Virginia|
|     Calvin Coolidge|1872-07-04|             Vermont|
+--------------------+----------+--------------------+
only showing top 20 rows

In the next post, we will see how to use LIKE operator to search for wildcard characters in Spark-SQL.

Leave a Reply

Your email address will not be published. Required fields are marked *