Practical Predictive Analytics

Issuing SQL through the R interface

Writing SQL this way consists of constructing a valid SQL string query via the sql function:

• The query must reference a registered table
• You can assign the results of the query to a dataframe:

Here is an example of getting a frequency count of the available samples from the registered object out_table. Note that the code begins with %r command which is a special Databricks directive that indicates that the code that follows applies to R. However, it is sometimes necessary to specify this (even in R notebooks), when a previous code chunks has used another language directive, such as %sql, or %python.

        %r         rm(tmp)         tmp <- SparkR::sql(sqlContext, "SELECT sample_bin,   count(*) FROM out_tbl group by sample_bin") ...