visto: an operational tool to visualise the duration of public investment projects
TRANSCRIPT
Copyright © 2009, SAS Institute Inc. All rights reserved.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
Carlo AmatiVISTO: An operational tool to visualise the duration of public investment projects
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
Outline
An introduction on public investments
VISTO: An operational tools for the interactive visualisation of project length estimates
VISTO behind the scenes: analytical insights
Examples
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
Public investments – What are we talking about?
Roadsand
motorways
Industrialareas
Watertreatment
plants
Airportterminals
Restorationof historical
buildings
Railwaysystems
Archaeologicalsites
AcqueductsUniversityfacilities
andmore
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
Public investments verification unit (UVER)
ProjectVerification
Audit Authority
On-site inspections
Assistance & support
Investment analysis
Monitoring and Statistics
Forecasting models
Monitoring systems
EffectivenessAppraisal
Assessment of results
UVER’s three areas
act as a bridge
between data and
policy
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
Public investments – Analysis framework
DataCollectionSystem
DataCollectionSystem
Data AnalysisSystem
Data AnalysisSystem
Data Interpretation
System
Data Interpretation
System
Relevant dimensions
Implementing parties
Implementing parties
Socio-economic context
Socio-economic context
Nature of project
Nature of project
Financialresources
ICT SYSTEMSICT SYSTEMS
Reality / Policy
achievements
Policy, rules and decision
making
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
The main stages of a project’s implementationPreliminary
DesignDetailed Design
Executive Design
Tender Works
The main purpose of the Preliminary Design is to define the dimensions of the structure, in order to obtain a clear picture of the structural action and make estimates of the project’s cost.
The main purpose of the Detailed Design is to define all the elements required to obtain the necessary permissions.
The main purpose of the Executive Design is to define all the details of the project in terms of costs and type of works.
The Tender stage refers to the period between the advertisement of the tender on public gazettes and the selection of the firm that will be contracted for the completion of the works.
The Works stage refers to the period between the handover of the site to the contractor and the completion of the works.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO at a glance
A tool for:
selection
benchmarking
early-warning
performance assessment
VISTO is:
simple in its interface but based on predictive models built on complex data
transparent to stakeholders and citizens
universal for a “clinical diagnosis” of investment projects
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: Application features VISTO yields an estimate of the time length of each
implementation stage of an investment project depending on its main attributes (cost, location, sector, type of project, category of implementing body, tender procedure).
VISTO yields the above estimates for each combination of attributes, regardless of whether such a project really exists or not.
Predictive models used for VISTO produce an estimate of the complete theoretical distribution of the time length of each stage: the median is chosen as central value and a variation interval is associated to it.
The length of each stage lasts until the start of the following one, except for the “Works” stage, ending at works completion.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: Panel description – Project features
The left-hand panel contains the project features that users can select interactively: the estimated length of each stage depends on the specific selected combination.
The available features are: amount, sector, type of project, category of implementing body, tender procedure, location at the provincial level.
The amount is the only quantitative feature and its value is typed in, all other features are selected with their combo box. Provinces are organised hierarchically and are selected within their region.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: Panel description – Length of stages
A button that produces a pdf output of the estimated lengths is available in the top right corner.
The upper part of the right-hand panel contains 5 dials where the estimated lengths of each of the 5 stages of a project with the selected features are represented together with some measures of the associated variability.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: Panel description – Dial elements
The black needle shows the estimated median value for the length of each stage, which is named “central length”.
The yellow band highlights a range of lengths, centred across the median value. The default width is set at 50% (25th-75th percentiles). The extremes are named “short” and “long” lengths.
The green sector marks the 35th percentile of the nationwide distribution (i.e. location is ignored) and can be used as a benchmark.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: Panel description – Overall project length
Below the summary chart, a green bar highlights the interval corresponding to the best 35% of projects with given features at national level, which may be considered as a benchmark.
The lower part of the right-hand panel contains a chart with the summary of the short, central and long lengths of the overall project, which are calculated as the sum of the lengths of the 5 stages.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: Potential users
A public officer can use VISTO as an additional tool for project selection during strategic planning of investments, or as an early-warning system in order to monitor the behaviour of on-going projects, or as an appraisal measure to evaluate the performance of a completed investment project.
A project manager can use VISTO to identify a reasonable target for his/her project, finding stimulus to push his administration towards the best practices at the national level.
A citizen can use VISTO to be aware on when a new public investment is most likely to be completed and available in his/her neighbourhood.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: Architecture
HTML version(compliance
with webaccessibilityregulation)
Results
Error page
Pdf output
Web serverMicrosoft IISbroker.exe (SAS)
SAS IntrNet
SAS Code
broker.exe
Application serverIBM AIXSAS 9.2 (+IntrNet)
CGI
SWF version
Model parameters
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: How it works – Homepage
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: How it works – SWF version
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: How it works – HTML version: Start page
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: How it works – HTML version: Results
...
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: How it works – HTML version: Error page
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: How it works – HTML version: Pdf output
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
Analytical insights
1. Qualitative information: classification through text-based clustering
2. From micro to macro: the importance of survival models
3. Does location matter? Analysis of regional effects and provincial dimensions
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
1. Text-based clustering – Categorical variablesProjects are not
always classified correctly in monitoring
databases, but the proper classification
may be recovered from other fields.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
1. Text-based clustering – Sectoral categorisation
………
……………
…………………..
……….
……….
………………….
……
……………….... and apply it to a
large number of cases.
“Restoration works of Trevi
fountain”
“Restoration works of Trevi
fountain”
The aim is to reproduce the categorisation process performed by the human mind ...
Works
ROADS
HOUSING
WATER SYSTEMS
CULTURAL HERITAGE
HOUSING
CULTURAL HERITAGE
…Restoration
TOURISM
CULTURAL HERITAGEFountain
TreviROADS
HOUSING
CULTURAL HERITAGE
…
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
1. Text-based clustering – The procedure (EM 5.2)
In Phase 1 data pre-processing is applied to a reference dataset where classification variables are known. It comprises the selection of project descriptions that can be used for the classification process (non-missing sector/type). Original description is enriched with sector description.Data are partitioned in a training and a test set (90-10).
Phase 1
Data pre-processing
PROJECT DESCRIPTIONS BY CATEGORY
DATA PARTITION
In Phase 2 the “textual” project descriptions of the reference dataset are transformed into a “numerical” space with the Text Miner Node. The Memory-Based Reasoning Node identifies similar cases by the k-NN algorithm and builds the classification model that can be applied to new cases.
Phase 2
Information processing
TEXT MINER MBR
In Phase 3 the Scoring Node allows the application of the classification model to new cases, i.e. project descriptions in the dataset that will be used for VISTO models.Classification is based on the proportion of “neighbours” of a new project description in each category (probability of classification).
Phase 3
Scoring
NEW PROJECTDESCRIPTIONS
SCORE
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
2. Survival models
Advantages
The analytic estimates are based on a model that takes into account completed stages as well as on-going ones (censored cases) Maximise the use of available information on stage lengths
Offer a complete estimate of the survival function, i.e. the probability that the length of a certain stage is exactly t, given the values of the descriptive variables
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
2. Survival models – FormulationLet Ti be a random variable representing the (possibly unobserved) survival time of the i-th unit. Since Ti must be non-negative, its logarithm can be modelled using a conventional linear model:
where εi is a suitable error term, with a distribution to be specifed. This model specifes the distribution of log-survival for the i-th unit as a simple shift of a standard or baseline distribution represented by the error term.
In SAS this can be easily coded as follows:
iinnii xxT ...log 110
proc lifereg data=ds;class x2 x3;model y*cens(1)= x1 x2 x3;
output out=res p=median;run;
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
2. Survival models – The role of censored data
The length of censored stages is the output estimate of the model.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
2. Survival models – The role of censored data
The width of the bars of the series “Only completed stages” is proportional to the ratio of “# completed” over “# completed and censored”.
The mean length of “Works” for the most costly projects increases by 20% when censored cases are included in the analysis. Their class (“More than 50 M€”) accounts for approx. 50% of the total financial value of all projects.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
3. Does location matter?
Two approaches:
Territorial representation with highest available detail, economic interpretation of effects:• Descriptive variables at the provincial level
+ dummy variables at the regional level
Measure of overall territorial efficiency• Dummy variables at the regional level
......log r rrk kk xxT
......log ' r rr xT :rx Regional dummies
:kx Provincial variables
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
3. Location matters: the effect of provincial vars
Variable (at the provincial level)
Economic Development Index 0.93 ** 0.95 * 1.09 ***
Environmental Condition Index 0.93 ** 1.05 ** 0.97 ***
Organized crime 0.86 *** 0.94 ** 1.05 ***
Crime aganist people 1.04 ***
exp(β ) exp(β )exp(β ) exp(β ) exp(β )
Tender WorksPreliminary
designDetaileddesign
Executive design
Better economic and environmental conditions are associated with faster completion of the first two stages, which mainly deal with administrative permissions, while are associated with longer time spent in defining the fundamental details of the projects: longer durations may indicate greater accuracy and efficiency.There is also a positive correlation between “Crime against people” and the length of the “Works” stage.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
3. Location matters: overall regional efficiency
0
1
2
3
4
5
6EMILIA-ROMAGNA
PIEMONTE E VALLE D'AOSTA
FRIULI-VENEZIA GIULIA
LOMBARDIA
MOLISE
VENETO
TRENTINO-ALTO ADIGE
ABRUZZO
MARCHEPUGLIASARDEGNA
CALABRIA
LAZIO
LIGURIA
TOSCANA
CAMPANIA
UMBRIA
SICILIA
BASILICATA
Regional mean for regional set Regional mean for national set National mean
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
3. Location matters: overall regional efficiency
0
1
2
3
4
5
6EMILIA-ROMAGNA
PIEMONTE E VALLE D'AOSTA
FRIULI-VENEZIA GIULIA
LOMBARDIA
MOLISE
VENETO
TRENTINO-ALTO ADIGE
ABRUZZO
MARCHEPUGLIASARDEGNA
CALABRIA
LAZIO
LIGURIA
TOSCANA
CAMPANIA
UMBRIA
SICILIA
BASILICATA
Regional mean for regional set Regional mean for national set National mean
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
3. Regionalefficiencyin otherwords
Very efficient (over 15% less than average)
Moderately efficient (3-15% less than average)
On average (±3%)Poorly efficient (3-15% more than average)Not efficient (over 15% more than average)
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
ReferencesFor further details on the above analytical insights:
1. C. Amati, F. De Angelis, F. Romani“Textual data classification for a sectoral categorisation of public investments” in Proceedings of Textual Data Statistical Analysis – JADT 2010, Rome, Italy.
2. C. Amati, F. Barbaro, C. Carlucci, F. De Angelis, A. Guerrizio“Survival analysis: an application to italian data on public investments” in Proceedings of Stochastic Modeling Techniques and Data Analysis International Conference – SMTDA 2010, Chania Crete, Greece.
3. C. Carlucci, A. Guerrizio, C. Amati, F. De Angelis, F. Barbaro “I tempi di attuazione delle opere pubbliche” in Proceedings of 30th Annual Scientific AISRe Conference – AISRe 2010, Aosta, Italy.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
VISTO: Where it is, and... why “VISTO”?
Soon available at:
http://www.dps.tesoro.it/VISTO
Working demo:
http://uver.altervista.org/VISTO_en.swf
Feedback:
VISTO, beside being an acronym, is also the Italian for “visa”: the idea is that a project whose stage lengths lie within the ranges indicated by VISTO can be acknowledged to specific benefits.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
Example 1: A completed project (MAXXI)
MAXXINational Museum of the XXI Century Arts
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
Example 1: MAXXI timesheet International design contest in 1998
Start of works in March 2003
End of works: 2009
Cost: 150 million euros
Overall length: 11 years
MAXXI:- falls beyond benchmark interval
- corresponds to the 40th percentile
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
Example 2: A new project (football stadium)
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
Example 2 (cont.): Hypothesis for an historical comparison
Started by Vespasian in 72 a.C.
Completed by Titus in 80 a.C.
With an overall implementation length of 8 years, Colosseum:
- would fall within benchmark interval- would correspond to the 28th percentile
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. S55547.0410
Thank you
web: www.dps.tesoro.it/UVERmail: [email protected]
Public investments verification unitDepartment for development and economic cohesionMinistry of Economic DevelopmentVia Liguria 2600187 Rome, Italy