| Author | Message | ||
     Nbkfan Kurra Bewarse Username: Nbkfan Post Number: 2003 Registered: 03-2006 Posted From: 14.100.132.10 Rating: N/A |
1. You need to have the data sets ready to run analytics on...raw data is not ideal...ex, impute, aggregate, transpose etc...these analytics ready datasets are referred as ABT 2. Random number generation techniques for some algorithms...ex ranuni 3. Whether u want to run algorithms in-memory or disk...based on that need to plan the servers and sizing appropriately... 4. Most importantly, variable selection techniques if ur datasets are huge and sparse...can use LASSO, Elastic Net etc 5. Preparing Validation and Test datasets 6. Deciding on algorithm techniques based on ur usecase scenario 7. Deployment options and opp planning...this is the key to success 8. Measurement and optimization of models over a period if time... More later... TIGER - SIMHA |