You should think of both the data and metadata folders and all of the composite XDF files stored in Hadoop in the 'data' folder as really all of the information for a single XDF output file. In Hadoop we create a set of composite XDF files(you will see files with name (*.xdfm) in the data folder. The metadata folder contains footer information for the XDF file such as number of variables, etc that is needed for fast file access.The reason we store XDF files in Hadoop this way is for fast file access and to take advantage of HDFS block sizes. You will need to store the results from rxPredict() as a composite XDf file in Hadoop or as a composite CSV file in HDFS.
When writing out predictions to XDF files in HDFS, why are results returned in metadata folder?
Need more help?
Want more options?
Explore subscription benefits, browse training courses, learn how to secure your device, and more.
Communities help you ask and answer questions, give feedback, and hear from experts with rich knowledge.
