-
Notifications
You must be signed in to change notification settings - Fork 0
Meeting Notes
Renzee Reyes edited this page Nov 9, 2018
·
3 revisions
- Modify BFNs to indicate insights; add direction to numbers
- Complete Midterm Presentation by next meeting
- Flesh out webpage mockup
- Renzee
- Scale down initial data. Less volume and only 3 months (Oct, Nov, Dec)
- Combine us-east and us-west to us
- Add more deviation between baseline
- Add insight - Europe and Asia ad campaign. Europe sees an increase in events but Asia not so much in February
- Frontload insights into last month for showcase in midterm presentation (ad campaign, style trend, issue with manufacturer slowing down packaging)
- Gaurav
- Take initial data to stage 2
- Focus on stage 1 for midterm presentation
- Establish stage insights
- In general
- Total Count by X
- At various time intervals
- Year/Several Months -
- Single Month -
- Week -
- Day
- In general
- Context
- New storefront with recently established e-commerce site
- Start from online order and take it to shipment
- Online order
- Purchase confirmation
- ...
- Check stockroom
- Package item
- Add e-commerce related fields
- Product Style
- User Demographic
- Get data concrete by Tuesday
- Meet again on Tuesday and Thursday
- Initial data and script pushed & committed
- Needed changes:
- Add unique ID
- Add event type
- Add origin
- Needed changes:
- Initial Tableau viz pushed & committed
- Simple connection of Tableau to data
- Sullivan to upload d3 templates
- Gaurav working on subsequent timestamp/stage generation script (By 10/26)
- Renzee to modify initial data to include:
- Unique ID
- Event Type ID
- Origin
- Node ID
- Low effort mockups in Tableau, Majority focus on d3
- Renzee - Finish initial data and template
- Use bootstrap for frontend layout
- Brush selection use cases
- More time-series data (messages per minute)
- Keep pipeline overview filtering simple; Checkboxes for inclusion/exclusion
- Start mockups in Tableau
- Draft specific insights; create our mockups around these insights
- Use public REST api for seed data
- Github repo setup for codebase (python, html, d3 js, etc): https://github.com/giktech/pipeline
- Gaurav to develop python script for generating staging (use numpy array function for specifying array of given mean and given std)
- Renzee to develop initial HTML environment (files, folders, css, libraries, etc)
- Research scenarios/context for various pipeline situations
- Retail (Amazon) - Spike in orders leading up to christmas
- Transit (Google) - https://developers.google.com/transit/
- Begin looking up examples to model our viz draft after (Sullivan to lead)
- Event Streaming Model (Kafka, Redis, HDFS, Spark)
- Ideal Meeting Time: Tuesdays @ 7:00pm PST (10:00pm EST, 4:00pm HST)
- Find a publicly available data stream