A major retailer was extensively using SAS for predictive analytics: There were more than 700 models built using SAS. The model building and variable generation code was taking more than 4/5 hours for each of the models and even longer for the models which were executed on larger and discrete data sets. The analytics team were unable to tweak the model frequently due to the time-intensive nature of the process.
Datametica migrated the models to run in memory on Hadoop where the datasets were first ingested into Hadoop HDFS. The model building code was replicated in H2O, R and Python. The variable generation and weightage calculation time were brought down to minutes.
Reduced cycle time for model building
Ability to tweak models when needed
More accurate predictions
Reduced errors with use of larger training data sets