This is an offshoot project of open source data quality (osDQ) project https://sourceforge.net/projects/dataquality/
This sub project will create apache spark based data pipeline where JSON based metadata (file) will be used to run data processing , data pipeline , data quality and data preparation and data modeling features for big data. This uses java API of apache spark. It can run in local mode also.
Get json example at https://github.com/arrahtech/osdq-spark
How to run
Unzip the zip file
Windows : java -cp .\lib\*;osdq-spark-0.0.1.jar org.arrah.framework.spark.run.TransformRunner -c .\example\samplerun.json
Mac UNIX
java -cp ./lib/*:./osdq-spark-0.0.1.jar org.arrah.framework.spark.run.TransformRunner -c ./example/samplerun.json
For those on windows, you need to have hadoop distribtion unzipped on local drive and HADOOP_HOME set. Also copy winutils.exe from here into HADOOP_HOME\bin
Features
- Create data pipeline like using Join, Filter, Aggregate, Case statement
- Use Data Quality - replace, drop, join,
- Data Profiling, Column base Profiling
- Fuzzy Join - cosine distance and others
- classification and sampling - random forest, Multi class neural network
- data normalization - zscore, std deviation, ratio score,
- Sampling Random, Stratified , Key based
Project Samples
