predictive modeling and reporting environment (pmre)
DESCRIPTION
Predictive Modeling And Reporting Environment (PMRE). CS 552 Senior Design Architecture Review. Presenting: Steve Su Ilya Chalyt Yuriy Stelmakh (Architect). Outline. MOV & Gantt Chart Architecture Sample Use Cases QFD & ICED-T Installation Procedure Function Points Questions?. MOV. - PowerPoint PPT PresentationTRANSCRIPT
Predictive Modeling And Reporting Environment (PMRE)
CS 552 Senior Design
Architecture Review
Presenting:Steve Su
Ilya ChalytYuriy Stelmakh (Architect)
Outline
MOV & Gantt Chart Architecture Sample Use Cases QFD & ICED-T Installation Procedure Function Points Questions?
MOV
PMRE will increase the efficiency and corresponding capacity of the USPS’s current parcel sorting infrastructure by 30% through predictive historical data analysis and strategic equipment placement while improving economic viability.
Engineering Management studies have shown that sorting machines run idle for approximately 30% of the day as employees search for bins by increasing the efficiency 30% (.30 x .70 = .21), to a new capacity of 91% leaving only 9% idle time. PMRE will minimize the amount of time these sorting machines remain idle.
Initial Gantt Chart
Current Gantt Chart
Architecture: Logical View Logical view of the system is based mostly on the
functional requirements. For PMRE the primary functional requirements are:
To predict future mail flow volume based on historical data Manage facility equipment including machines and bins Manage users and operating shifts
Architecture: Logical View
Equipment Management
Staff Management
Analysis and Predictions
Data Services
Architecture: Logical View
Equipment Management Class Category will consist of three classes:
Machine Management Class Locations Management Class Bin Management Class
Equipment Management
Machine Management
Class
Bin Management Class
Locations Management
Class
Architecture: Logical View Staff Management Class Category will consist of
two classes: User Management Class Shift Management Class
Staff Management
User Management Class
Shift Management Class
Architecture: Logical View Analysis and Predictions Class Category will
consist of two classes: Data Analysis Class Predictions Class
Analysis and Predictions
Data Analysis Class
Predictions Class
Architecture: Process View
At highest level of abstraction PMRE will have 4 distinctive processes: Data Storage and Collection Process Analysis and Prediction Process Notifications Process RFID Tracking Process
Architecture: Process View
Analysis and PredictionProcess
NotificationsProcessRFID Tracking
Process
Data Storage and Collection
Process
Architecture: Process View
Now let’s look at the Data Storage and Collection Process. It will consist of four main processes:
Login Process Data Collection Process Data Transfer Process Logging process
Data Storage and Collection Process
Login Process
Data Collection Process
Data Transfer Process
Data Logging Process
Architecture: Physical View
PMRE server will be running Apache 2.0 with PHP 4.3 and MySQL 4.1 database
Client PC requirements are the following: IE 5.0+ Firefox 1.0+ Netscape 7+ OS capable of running any of the above
INTERNETClient
Client
Client
PMRE SERVER
USPS INTERNAL NETWORK
EXTERNAL CLIENTS
Architecture: Development View
PMRE has three distinct development layers as follows: Database Layer API Layer UI Layer
Database Layer Database Connection Simple Data Queries Data Access Levels
API Layer PHP functions Equipment Management Staff Management Analysis And Predictions
User Interface All UI Pages PHP and HTML
Software Engineering Principles
WinWin Spiral Model Ensures that all parties benefit and are considered in
the goals of the application. Change Freeze
Stabilizes requirements Allows for progress to be made which may help to
alleviate the requirements confusion. Wideband Delphi Methodology
One or two experts create a specification which is then reviewed by a panel of members.
Complete Use Cases
Main Page
/ System Access
/ Click HelpHelp Page
Incorrect Login Page/ Incorrect Login
/ Click Home
Main Access Page
/ Correct Login
Equipment Management Page
Statistics Page
Shifts Page
Locations Page
Bin Types Page
User Administration Page
Data Input Page
/ Click Equipment Management
/ Click Statistics
/ Click Shifts
/ Click Locations
/ Click Bin Types
/ Click User Administration
/ Clck Data
/ Add A Bin,Remove A Bin,Change Bin State
/ Set Recomendation,Set Algorithm
/ Add A Shift,Remove A Shift,Modify Shift Start Time,Modify Shift End Time
/ Add Location,Remove Location,Modify Location Owner,Modify Location Type,Modify Location Quantity,Modify Location Description
/ Add Bin,Remove Bin,Modify Type Volume
/ Add a User,Remove a User,Set Time
/ Modify Data Point,Delete Data Point,Input Data Point
Inadeqate Perm
issions Page
/ Click Back To Main Access
User Logon and Main Menu
Main Page/ System Access
/ Click HelpHelp Page
Incorrect Login Page/ Incorrect Login
/ Click Home
Main Access Page
/ Correct Login
Equipment Management
Main Access Page
Equipment Management Page/ Click Equipment Management
/ Remove A Bin,Add A Bin,Change Bin State
Inadeqate Permissions Page
/ Click Back To Main Access
Shifts Page
Main Access Page
/ Click Shifts
Inadeqate Permissions Page
/ Click Back To Main Access
Shifts Page
/ Modify Shift Start Time,Add A Shift,Remove A Shift,Modify Shift End Time
Statistics Page
Main Access Page
/ Click Shifts
Inadeqate Permissions Page
/ Click Back To Main Access
Shifts Page
/ Modify Shift Start Time,Add A Shift,Remove A Shift,Modify Shift End Time
Bin Types Page
Bin Types Page
/ Remove Bin,Add Bin,Modify Type Volume
Main Access Page
/ Click Bin Types
Inadeqate Permissions Page
/ Click Back To Main Access
User Administration Page
User Administration Page
/ Add a User,Remove a User,Set Time
Main Access Page
/ Click User Adminstation
Inadeqate Permissions Page
/ Click Back To Main Access
Data Input Page
Data Input Page
/ Modify Data Point,Delete Data Point,Input Data Point
Main Access Page
/ Click Data
Inadeqate Permissions Page
/ Click Back To Main Access
Locations Page
Locations Page
/ Add Location,Modify Location Type,Modify Location Quantity,Modify Location Description,Modify Location Owner,Remove Location
Main Access Page
/ Click Data
Inadeqate Permissions Page
/ Click Back To Main Access
QFD – House of Quality Matrix The QFD matrix helps
details the understanding of the customers needs and wants.
The House of Quality Matrix allows us to make a connection between system functionality and the technology implementation.
Based on customer requirements and technical solutions, a rating of the implementation solution is created. In our case, system administration and data availability are ranked as being the most important.
●
Customer Rating Poor Good 1 2 3 4 5
System makes recommendations on quantity and location of bins 6 ● ● ○ ● √All users able to view a read-only snapshot of the facility 5 ○ ● ● ● √All users able to view historical read-only snapshots of the facility 6 ○ ● ● √System users are able to log-in and administer the PMRE system 9 ○ ● ● √System users are able to configure staging areas 6 ○ ● ● √System users are able o configure shifts and equipment 7 ○ ● ● ● √System users are able to generate predictions for bin use in sorting facility 5 ● ○ ● √System users are able to configure system prediction algorithm 5 ● ○ √System users are able to view and post mail volume data. 4 ● ● ○ √System compares recommendations vs. actual measured activity. 5 ● √
1 2 3 4 5 63 4 2 1 5 6
Priority measured on 1 to 9 scale, 9 being the most important. ● Strong Relationship
○ Weak Relationship
X Conflict
Dat
abas
e (m
ySQ
L)
Pro
gram
min
g La
ngua
ge [P
HP
]
●○
QFD - House of Quality Matrix
Difficulty
Web
Ser
ver
Inte
rnet
App
licat
ion
Web
Use
r In
terp
hase
Importance
Prio
rity
(wei
ght)
Web
Adm
inis
trat
or
Inte
rpha
se
Implementation Ease vs. Feature Importance
• The Implementation Ease vs. Feature Importance breakdown allows us to rank and determine which aspects of the project will require the most time and effort.
Generating and Viewing Predictions
Administrating the
users/system
Configuring the equipment
& shiftsFacility read-
only snapshots Total
Fast user interface 7 5 5 3 20Reliable application 9 5 7 5 26Correct data generation 9 3 9 7 28Availability of the data 7 7 7 5 26Totals 32 20 28 20
1 Weak3 Moderate5 Strong7 Very Strong9 Extremely Strong
Implementation Ease vs. Feature Importance
ICED-T Metrics
ICED-T Rating
1 The application UI is easy to use 42 Modifying Shift / Equipment 43 Generating resoults 5
1 Predictions made are consistent from day to day 32 Administrator can control and modify the users 5
1 The time to generate a report 42 The overall snapshot of the fecility is available on the home pae 4
1 The application is available from mulitiple locations 52 Supports multiple users and protects againsts modifications. 4
1 The user doesn't need extensive training to use the product 42 Normal user actions and task can be easily performed 5
Thoughtful
Intuitive
Consistent
Efficient
Durable
(3) Same as other applications I have used(4) Better than average(5) Best I have ever seen
ICED-T Metric
ICED-T Model Rating System(1) Worst I have ever seen(2) Worst than average
Installation Procedure
Microsoft MSI Installer
Run the installer Restart the Computer
Installation Procedure
Advanced Installer – From Source1. Install Apache using the standard Apache Runtimes 2. Install PHP 4.3+ using the standard PHP Runtimes 3. Install MySQL 4.1+ using the standard MySQL Runtimes 4. Copy the provided PMRE Source to the WWW Root / World
Readable directory 5. Set the proper port forwarding as per your companies
specific infrastructure 6. Populate the MySQL Database using the supplied
INSTALL.sql file and the MySQL Batch loader 7. Restart the machine or respective services
Installation Procedure
RFID Tracking Station Install
1. Install Microsoft .NET 1.1 and 2.0 using the standard .NET Runtimes
2. Install the RFID Tracker from the provided installer
3. Set the MySQL Server address, port, and login information
4. Restart the machine
Unadjusted Function Points
Component Count Complexity Product
Inputs 3 6 18
Outputs 1 4 4
Internal Data Files 1 10 10
External Interfaces 0 0 0
Inquires 1 6 6
Total UFPs 38
By analyzing the structure of our application we can make critical staffing decisions and plan for complexity related issues that are now expected.
Technology Complexity FactorData Communications 5 SSL communications
Distributed Functions 0 No distributed processing
Performance 4 Users should have feel of quick response
Heavily Used Configuration 1 Simple configuration
Transaction Rate 4 Built to handle large amounts of simultaneous users
On-line Data Entry 5 Complex Web Interface
End-User Efficiency 5 Interface must be efficient for user use
Online Update 2 Some components require online updating
Complex Processing 3 Recommendation algorithm
Reusability 5 Must be maintained for RFID and other evolutionary requirements
Installation Ease 5 Must be easy to configure and install
Operational ease 3 Database backup and recovery important
Multiple Sites 5 Must perform identically from many sites.
Facilitation of Change 1 Static historical data
VAF 48
Adjusted Functional Points and Effort Estimate
Give a staff of 10 personnel for 10 month, working half time with a fifty percent reduction in productivity due to added communications costs yields a total of 10*10*(1/2)*(1/2) = 25 effective staff months, 7.26 times the necessary project staffing. This project is over staffed.
AFP=UFP(.65+.01*VAF) AFP=38*(.65+.01*48)= 42.94 PHP is most similar to PERL, for which
Quantitative Software Management reports SLOC/FP of 60.
LOC = 43 * 60 = 2580 Productivity: 2580LOC -> 750 LOC per month Effort: 2580 / 750 = 3.44 Staff Months
Net LOC per Staff Month
0
500
1000
1500
2000
2500
1 10 100 1000 10000 100000 1000000
Size of Program in LOC
Net
LO
C p
er M
on
th
Thank You
Questions?