Shortcut to Parametric Estimation via Web Trnsport

1
Shortcut to Parametric Estimationvia Web
Trnsport

• Charles Engelke
• Info Tech, Inc.

2
The Parametric Paradox

• Parametric estimation is one of the most
  important and desired estimation techniques
• That very few people use

3
Why Dont More People Use Parametric Estimation?

• Because it requires powerful software?
• No, Trnsport CES is available

4
What Most CES Users See
5
The Real Obstacles

• Requires lots of resources and information
• The scope and breadth of issues is very large
• Methodology is complex, requires a lot of
  training and effort to master
• A lot of fundamental data must be gathered and
  managed before the first estimate can be built
• These are the big issues for CES users

6
Parametric Estimation is More Varied than Detail
Estimation

• The models for estimating an item price are the
  same regardless of the item
• Cost-based
• Determine crew, productivity, materials
  consumption, overhead, margin
• Plug in current costs and derive unit price
• Bid-based
• Calculate weighted average prices or fit to
  standard formula via regression
• Apply results to predict new item bid prices

7
Parametric Estimation Models Vary for Different
Job Types

• Steps needed
• Develop conceptual model
• Define the model as a template for new jobs of
  that type
• Define how bid items are related to cost drivers
• Statistically analyze historical prices
• Enter model parameters and calculate
• Only the last two steps apply to detailed
  estimation with CES or Estimator

8
The Real CES Challenges

• New CES users start with a blank screen for the
  parametric model templates
• Cant develop estimates until all those
  preliminary definitions are done
• Its hard to see how to specify the abstract
  model so it can be used for real jobs
• And there is always other, more urgent, work to
  be done

9
The Fundamental Dilemma

• CES is extremely flexible
• Any parametric model that can be reduced to
  formulas of any kind can be defined in it
• But the first steps are extremely steep
• And its hard to understand these steps without
  experience using other models
• The necessary flexibility and power in the system
  lead to barriers to entry

10
Simplify, simplify.

• Parametric estimation is very complex and
  difficult in general
• So forget the general problem for now
• All of CES will still be there to handle it when
  we come back!
• Pick one type of problem, and create an express
  lane for it in CES
• Make a shortcut to parametric estimation

11
The Shortcut(s)

• Choose specific types of estimates to support
• Asphalt resurfacing, perhaps another
• Pick one estimation model per type
• That is, determine what parameters will be
  considered, and how estimates will be derived
  from them
• Provide reasonable defaults for everything
• So estimators can provide minimal data and get
  adequate results, or more data for better results
• Use defaults to simplify installation, too

12
Example Resurfacing

• Minimal parameters
• Length, lanes, lane width, depth
• Optional parameters
• Location, urban/rural, maintenance of traffic,
  mobilization, milling, drainage, earthwork
• Underlying data for calculations
• Acquire cost data, and classify and analyze
  historical data

13
Shortcut Work Flow (for the user)

• User enters required parameters
• An estimate total is provided
• User modifies defaulted parameters
• The estimate is adjusted automatically
• User wants to change basic estimation model
• Must leave the shortcut and continue work in full
  CES instead

14
Shortcut Work (for the agency)

• Install shortcut, accept default data
• Get reasonable estimates, but not based on
  anything specific to your agency
• Classify basic historical data
• Define a few item groups and conversions
• Run analysis models to feed CES
• Develop your own models
• Acquire cost driver data specific to your issues
• Again, must leave shortcut for full CES at some
  point

15
Challenges for a Shortcut

• We dont want to add it to todays CES client
• Thats already big enough
• Trying to make something simple by adding to a
  complex product isnt likely to succeed
• We dont want to create a new application
• Those are expensive to build and maintain
• It might be the first of many specialized
  applications, which would be hard to manage
• What options are left?

16
Enter Web Trnsport

• An initiative to replace use of the Trnsport
  Windows clients for some functions
• More ambitious than Trnsport Intranet
• Not just a supplement to client-server
• Web functions are inherently very loosely coupled
• So specialized functions can be developed
  independently of others
• Within a single manageable framework

17
Web Trnsport is not a New Application
Only the blue pieces are new
18
Strategic Direction to Web

• Lots of reasons for Web Trnsport
• (Ill gladly harangue anyone wholl listen)
• Part of strategic plan for several years
• So Info Tech proposed a project to create needed
  infrastructure
• And the TTF said whats in it for the users?
• Every project should provide direct user value
• Project redirected to start Web Trnsport to
  solve important user problems
• Starting with parametric estimation shortcuts

19
Project Goals

• Provide a useful function almost all licensees
  will want to deploy
• Parametric estimation shortcuts
• Be easy for licensees to implement quickly
• We dont want to deliver shelfware
• Ready for people to use almost immediately
• Without a lot of configuration, data management,
  or end-user training
• Create architectural foundation for Web Trnsport
  while delivering immediate value

20
Project in a Nutshell

• Performing this fiscal year
• Late start due to redirection of goals
• Using agile methodologies
• Replaces traditional lifecycle (with several
  review and approval stages) with constant
  communications
• Product builds weekly or more frequently,
  available instantly to TRT
• Delivery Q3 2004

21
What Were Shooting For
22
Issues

• Flexibility versus simplicity tradeoffs
• Where do we draw the line?
• Integration with agency data
• Can we make matching items to cost groups as
  simple as the rest?
• Authorized users
• Same as for similar CES screens, or a broader
  group?
• Allow guest estimators? Planners could create
  rough estimates, save in CES for estimators to
  develop more fully later.

23
What Can You Do?

• Communicate with the TRT
• Agile methodologies mean more functional
  decisions happen during the project
• Discuss what you want
• Chat about it on the TEA mailing list
• Get your system managers ready!
• A successful project will result in something
  quick to install and deploy

24
Shortcut to Parametric Estimationvia Web
Trnsport

• Charles Engelke
• Info Tech, Inc.