[ARROW-9218] [R] Numeric columns turn to string when imported in R

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Not A Problem
• Affects Version/s: 0.17.1
• Component/s: R
• External issue URL: https://github.com/apache/arrow/issues/25317
• Assignee: Unassigned
• Reporter: David Cortes
• Created: 24/Jun/20 14:27
• Updated: 11/Jan/23 08:05
• Resolved: 26/Jun/20 10:23

Description

If I create a snappy-compressed parquet file from a DataFrame in spark or pandas, and then import this same file into R using:

arrow::read_parquet(the_file, as_data_frame=TRUE)

or

arrow::read_parquet(the_file, as_data_frame=FALSE)

Then the columns that were numeric/float before will load as strings.

Loading the same file in Python through

pd.read_parquet(the_file)

Will correctly interpret numeric columns as numeric.

 

Integer columns seem to be read as integers however.