Search code examples
scaladataframeapache-sparkrdd

Count occurrences in dataframe of arrays

I have a dataframe that looks like this:

|--------------------------|
|          text            |     
|--------------------------|
|  ["hello", "world"]      |
|--------------------------|
|  ["foo"]                 |
|--------------------------|
|  ["world", "bar"]        |
|--------------------------|
|  ["kung", "foo", "world"]|
|--------------------------|
|          ...             |

And I need to count the occurrences of each word and sort it from greatest to least. I do not know all the words it may contain. How can I manipulate this to do that?

It would look something like this when done

|-----------------|-----------------|
|        word     |       count     |
|-----------------|-----------------|
|     hello       |         1       |
|-----------------|-----------------|
|     world       |         3       |
|-----------------|-----------------|
|     bar         |         1       |
|-----------------|-----------------|
|     foo         |         2       |
|-----------------|-----------------|
|     ...         |        ...      |

I've tried flattening the dataframe, maping over it, etc. etc. and I have super stuck. Any help would be appreciated!

Solution

  • Presuming that has been stored as a list of string:

    df.select(explode($"text").as("word")).groupBy("word").count.orderBy(desc("count"))