PROJECT DATA FLOW IS IS AN ENGINEERED SYSTEM PLANNING YOUR DATA SYSTEM TO REACH YOUR GOALS KEN JOHNSON NASA ENGINEERING AND SAFETY CENTER/ NESC INTEGRATION OFFICE NASA STATISTICAL ENGINEERING TEAM APRIL 5, 2017 R170329
HERE’S YOUR DATA Project Data System Planning April 5, 2017 2
WHAT’S HERE • Problem statement for this presentation: • Want to make sure the structure and process of passing information through a project is recognized for what it is: A SYSTEM. • Show how system planning tools can make information interfaces preserve/ add value and focused on SOLVING THE PROBLEM. • Goal of this presentation • Leave this room expecting to plan your next project data system through win-win negotiation between data supplier and data customer Project Data System Planning April 5, 2017 3
PROJECT DATA FLOW PER TEST, 100 TESTS Pre-Test Drawing Analog Data (dimensional Acquisition Systems data) Strain Fields Pdf of picture of drawing from Imaging showing dimensions Test High-Speed Post-Test Drawing Requirements Videos Verbal (dimensional Pdf data) Info Oscilloscopes Pdf of same Flash Report Large-Data Flow drawing Timing Camera Spec (another diagram) with ink pen Word Diagram Sheet markups File Many Pdf of Excel Excel Files and Worksheet File File Storage Formats Problem statement: Everything This is inefficient. Analysis Project Data System Planning April 5, 2017 4 We want less waste.
WHAT’S DATA? WHERE’S IT GOING? Data is any kind of raw information that flows through a project Post-Test Drawing Generated and gathered for a purpose: ANALYSIS (dimensional data) • Quantitative Pdf of • Qualitative drawing • Numbers with ink pen markups • Images File Storage • Configuration info • Dimensions • Instructions Analysis • Drawings – specs – models – …….. Project Data System Planning April 5, 2017 5
THE SIMPLE SYSTEM TRANSACTION MODEL Post-Test Drawing (dimensional data) Pdf of Pdf of drawing Drawing drawing with ink pen Analysis Dims with ink pen markups markups Analysis Project Data System Planning April 5, 2017 6
THE SIMPLE SYSTEM TRANSACTION MODEL Pdf of • Information products flow from a Drawing drawing Analysis Dims with ink pen data supplier to a data customer markups Project Data System Planning April 5, 2017 7
THE IDEA • Think of this as a system • The grey arrow represents a Pdf of system interface Information drawing Drawing Data Data Analysis • … showing a transaction Supplier Dims with ink pen Products Customer markups between subsystems • The transaction is carried out using data as the product Project Data System Planning April 5, 2017 8
GOOD NEWS: WE HAVE SMART HUMANS IN THE LOOP • Subsystems generally currently include people who can negotiate directly to plan an optimized Information Data Another Data Data Engineer transaction and interface Engineer Supplier Products Customer • No competition here: all negotiations are win-win Project Data System Planning April 5, 2017 9
SYSTEM INTERFACE/ TRANSACTION QUALITY Effective Interfaces Poor Interfaces • • Test: can the data move from source directly into the Test: does the data user have to do something to be able intended analysis engine without unnecessary work? to analyze the data? • • All in one place Scattered • • Electronic spreadsheet/ database Multiple pages, documents, formats, even sites • • Easy to integrate disparate data (key fields) Hard to relate one dataset to another (config control) • • NOTES in their own fields integrated into the database Verbal information • • Standard rules followed Unanalyzable format: pdf, handwritten, graphs, summaries • Row-column format • Pretty • One piece of information per cell, one kind of • information per column… Information stored in formatting but not in cells • All fields necessary to reach goals are present • Key information … where? • • Negotiated between data supplier and analyst Specified by the data supplier Project Data System Planning April 5, 2017 10
ANALYZABLE DATA TABLE FORMAT • Standards such as ADaM (see backup), IEEE, … • Easy and basic example: UCLA Institute for Digital Research and Education http://stats.idre.ucla.edu/other/mult-pkg/faq/general/tips-for-creating-an- excel-file-that-can-be-easily-moved-to-a-statistical-program-for-analysis/ • Uses a particular analysis program, but easy to get the idea – and links to an example of a lousy data table • High value? Talk to an expert • Data Mining Team (DMT/ Bob Beil ) … Project Data System Planning April 5, 2017 11
PLANNING A SUCCESSFUL TRANSACTION: FOCUS ON SOLVING THE PROBLEM To solve the What data problem, I need: do you need 1. Location to solve the 2. Dimension problem? • Goals of the transaction 3. … 1. Make sure the final customer gets what s/he needs: a solved Data Data Data problem Supplier Customer 2. See Goal #1 Project Data System Planning April 5, 2017 12
DATA SUPPLIER’S RULE #1: EXPECT YOUR DATA TO BE ANALYZED .csv is better. Be The data recorder • Deliver data in analyzable format sure that column can give you output in .csv or headers are in • Find out how the analysis will be Excel format. exactly one row. performed • Set system/ transaction requirements Data Data Data Supplier Customer • … negotiate/ iterate as necessary • Don’t make extra work • … for supplier OR customer Project Data System Planning April 5, 2017 13
DATA SUPPLIER’S RULE #2: TAKE IT EASY Can you set up the data How are we • KISS BNTS – keep the interface system recorder so it outputs handling simple (but not TOO simple)! test date, test number configuration and test article number • Record it all control? as data columns? • Columns are cheap ; cleaning and reconstructing data are expensive • Consider delivering raw data with reduced data (ask customers) Data Data Data • Pass all important information through the Supplier Customer same interface • Consider recording data in the finished form right at the source • Make the machine do it Project Data System Planning April 5, 2017 14
DATA CUSTOMER’S RULE #1: COMMUNICATE WHAT YOU NEED • Have an analysis plan focused on Glad you asked. What do you Let’s grab lunch need? solving the problem and talk about it. • Know what inputs are necessary for the analysis • Know the format needed for Data Data Data Supplier Customer analysis • Communicate this to your supplier Project Data System Planning April 5, 2017 15
DATA CUSTOMER’S RULE #2: REMEMBER: YOU’RE A SUPPLIER, TOO Glad you asked. I’ll get What do you Let’s grab lunch need? my coat! • Remember the goals of your and talk about it. analysis – and of the overall task • Negotiate concurrently with your Data customers and suppliers Data Data Data Customer Supplier Cu • Better yet: negotiate together as a / Supplier system Project Data System Planning April 5, 2017 16
ADD VALUE … and my analysis The tech just enters • Build in visualizations software reads the the data into form fields right on the data directly off • This is where you do pretty drawing… the same database. • Report summaries, graphs, … • Consider real-time visualizations Standard- Pdf of Drawing Format with Data drawing Analysis Drawing Data Data • Pass data forward in standard Data Analysis Supplier Entry Dims with ink pen Relational* Customer Software analyzable format for future users Fields Database markups • Give future projects a database to add to Project Data System Planning April 5, 2017 17 * Look it up. You want one of these.
WHAT WE NEED TO DO: SYSTEM AND STATISTICAL ENGINEERING • Think about the future • Problem statement (ALWAYS) • Expen$ive data • Goals • Future customers are customers • Plan for success • Longevity • Expect to analyze data • Complete • Data transfer in a format that makes sense • Well-documented • Negotiate between supplier and customer • Useful • Think about the system • Efforts • How can waste be pulled out of the system? • NESC’s Data Mining Team (DMT – Bob Beil) • How can value be added? • Big Data Working Group … Project Data System Planning April 5, 2017 18
THANK YOU THANKS TO BOB BEIL (NESC/ KSC), JILL PRINCE (NESC/ LARC), CHRIS KOSTYK (AFRC-RS), JON HOLLADAY (NESC/ GSFC), VICKI REGENIE (NESC/ AFRC) Project Data System Planning April 5, 2017 19
BACKUP Project Data System Planning April 5, 2017 20
TMI: FUNDAMENTAL PRINCIPLES OF THE ANALYSIS DATA MODEL (ADaM) Clinical Data Interchange Standards Consortium (CDISC) Analysis Data Model Version 2.1 https://www.cdisc.org/system/files/members/standard/foundational/adam/analysis_data_model_v2.1.pdf Project Data System Planning April 5, 2017 21
Recommend
More recommend
