#pdltools

Accomplished:

set up dual monitors

finished code for free fall customer id compilation

unique customer id count has decreased...may be due to month purge of underlying base table (unique customer bridge table)

started exploring attribute tables (booking, purchases)

Notes:

PDL Tools usage

@delimiter is needed in order to run the sugarlib function from DBVisualizer. See http://confluence.dbvis.com/display/UG95/Executing+Complex+Statements#ExecutingComplexStatements-UsinganSQLBlock

added two more business units to scope of purchase order data (170, 172)

decided to filter booking data to

curent_status_ind = 1 and order_line_status=11 or 12 --> customer's booked and checkin part

curent_status_ind = 1 andorder_line_status= 4 --> customer's booked part

order_status_id values repesent status of booking or purchase ie. Booked, Cancelled, past visit, returned, etc

current_status_ind indicates if it is the current status of the order

Accomplished:

Identified a bug in the present data compilation process and fixed it

Finished optimizing the data compilation code

Showed Quan how the new optimized code cleaner and faster

Paired with Quan to start building Logistic Regression model in MADlib

PDLtools is still a bottleneck, contingency plan is to just copy past code from pdltools github

Can bring up pdltools in Amy's meeting as its probably her team installing things on the CDE

Overview of the progress so far for Amy: First Week: - Investigated ways to integrate SAS models to score in GPDB, - Implemented scoring of the free-fall model while pairing with Victor - Walkthrough of various model workflows with Quan and Deepak

Second Week: - Presented the best practices & recommendations deck to Quan, Deepak and Victor - Deep dive into the frozen free fall model from Quan - Setting up logistics of network access, db access and required tools