Project Management VII Research Paper

1University of Sydney, New South Wales, Australia

Corresponding Author: Michael Hopmere, University of Sydney, New South Wales, 2006, Australia. Email: mhop9461@ uni. sydney. edu. au

Project Management Journal 2020, Vol. 51(6) 656–669

© 2020 Project Management Institute, Inc. Article reuse guidelines:

sagepub. com/ journals- permissions DOI: 10. 1177/ 8756 9728 20933446

journals. sagepub. com/ home/ pmx

Proactively Monitoring Large Project Portfolios

Michael Hopmere1, Lynn Crawford1, and Michael S. Harré1

Abstract The discipline of project management has evolved over the years, yet projects still run into trouble, failing entirely, running late, or not delivering expected benefits. Program and portfolio managers need assistance identifying potentially troubled projects while they are being delivered, allowing time to intervene. We report on our investigation of whether project status reports from IT project portfolios can be used to predict projects that may be trending into trouble ahead of time. We found that this initial approach resulted in a high degree of accurate predictions opening new avenues of research in predicting project progress and health.

Keywords project portfolio management, risk management, early warning signs, time series forecasting

Article

Introduction The discipline of project management has evolved greatly over the years, yet projects still run into trouble—failing entirely,

thousands of projects underway at any given time, program and portfolio managers need assistance identifying potentially trou- bled projects while they are still being delivered, allowing them time to intervene to correct project health issues while there is

Referring to the repeating historical phenomena of what he

changes, which threaten a reversal or loss of a major opportu- nity) that befall organizations,

problem before the fact and thereby minimize the probability of

This work focuses on the second approach above, describing a before the fact forecasting system to identify emerging proj- ect issues early thereby reducing the suddenness and urgency, which may allow an organization to address these threats pro- actively (

The main contribution of this work is our investigation into the question of whether it is possible to forecast project trouble over an actionable future horizon using project status reports

upon the use of data to inform decision making and improve project performance, this research also complements increas- ing recognition of the critical role of digital skills and data ana- lytics in the future of projects and project work (Project

, From a theoretical perspective, this work is founded on proj-

ect risk management, which is focused on limiting the negative impact of uncertain events and reducing the probability that these events will occur ( we describe a system that can continually monitor project per- formance information across an entire portfolio, identifying risk triggers or early warning signs of emergent risks that may

project portfolio management toward “proactive monitoring” by presenting an approach to accurately predict the future

early warning signs research by using a very large data set cov- ering thousands of projects from a single industry and technol- ogy area, and by analyzing data collected live during the

and beyond the

Hopmere et al. 657

contributions of and by using large amounts of data collected continuously

over the life of a project for the purposes of establishing a “before the fact” ( ) forecasting system that allows more time for project management practitioners to react to

Lastly, we contribute in a limited way to the research area of project complexity by applying an analytical approach to proj- ect management data that is akin to those that might be applied

are following the lead of whose work establishes that projects could indeed be considered to be com- plex systems, opening the opportunity for ourselves and other researchers to bring to bear the tools and approaches well

Literature Review Risk Management and Early Warning Signs: It is well known that many projects run into trouble or fail ( ; ;

; ; ; ) Project Management Institute (PMI)

Much has been written about the causes of project failure ( ; ; Hughes

; ; consider that project failure can only be assessed after a project has been completed, allowing time to measure the degree to

failure is essentially locked once the project is completed, leav- ing no levers for the delivering team to correct any undesired

project trouble to be the

customer satisfaction concerns, and many other issues experi-

failure, with relevant information and skill, project trouble can

ranging causes of project failure, rather we focus on forecasting

There has been considerable research on the early warning signs of project trouble or failure ( ; Geller

; ; ; ;

; , ; early warning signs is not

unique to project management, but within project management, -

course ( ; ; ; ,

established that per theory of weak signals, project problems did indeed appear

as preliminary signals that could be leveraged in project and portfo- lio management, and more recent analysis of post- mortem reports for unsuccessful IT projects indicated that early warning signs can be picked up in a project (

Early warning signs, sometimes called risk symptoms (PMI, ) or risk triggers (

event or situation that indicates that a risk is about to occur” (

weak signal as being an imprecise early indication about an impending impact- ful event ( ; Nikander & Eloranta,

being an omen, or an indication of future developments ( consider these terms to be interchangable, sharing roots in the

In project management, the term risk “an uncertain event or condition that, if it occurs, has a positive

-

is to sort risks based on the degree of available

organization is aware of an uncertain event), unknown- knowns (these are hidden facts, where some knowledge exists, but the organization may not be aware of it at the time of the endeavor [ ]), and unknown- unknowns (these are emergent risks that are essentially unknowable unless they begin to manifest [

states that project risks are a collection of alea- toric risks (those to which probabilities can be objectively assigned—which would include known- knowns and known- unknowns) and epistemic risks (those stemming from a lack of knowledge—including unknown- knowns and unknown-

is referring to when he mentions the occurrence of strategic surprises (an unex- pected discontinuity or deviation from the expected future trajec- tory for a given product or situation ( ; Haji- Kazemi

that for emergent risks, that is, risks that could not have been identi- -

also notes that the risk management meth- ods used in practice generally focus on the monitoring and

risks or known- unkowns), which, despite the guidance that project, program, and portfolio managers should continuously monitor their projects for the emergence of previously uniden-

; , ;

Project Management Journal 51(6)658

- Cavallo & Ireland,

; ), there is a growing concern that this approach does not adequately take into account the combined

Cavallo & Ireland, ; ;

are often caused by multiple problems that were not or could not be predicted earlier in the delivery of a project, and these problems may cascade (penetrate into multiple adjacent sub-

( ), which may amplify the impact of these

concurrent risks as early as possible in the project life cycle a critically important area in project management (Thamhain,

Risk management has been one of the core knowledge areas in project management for many decades (Petit & Hobbs,

- tive impact of uncertain events and/or reducing the probability of these negative events materializing ( The risk management literature is large, but in general it is agreed that the project risk management process includes activ- ities similar to the following ( -

risks (

surveillance of risk trigger conditions (or early warning signs), should be a continuous process throughout the project (

; ;

quite generic in nature, for example, ”requirements instability” ( ) or “weak project manager” (

team, like “immature technology” ( ) or “lack of top management support” ( ),

-

to identify causes for project failure and associated early warn- ing signs are often collected after the subject projects have been completed ( ; ; Klakegg

; ; ) and generally from a small set of very large case study projects ( ; ; ) or from small sample groups of respondents for a given project (Keil,

Project Management and Status Reporting Projects are an important organizational construct used to plan and control the delivery of a vast array of products and services ( ; ;

; -

considering all project types and industries, it could be reason- ably argued that there are many tens of thousands of projects

Project status reporting is a common tool for communicating with key stakeholders, providing a quick view of progress, plans,

Process Group ( ), a periodic (often weekly or monthly) status report is used to summarize project information (Haji-

; ; ; ) as

and future outlook ( ; To aid reporting and monitoring across concurrent projects,

various display techniques have been employed by project

simplify and combine large amounts of information into a com- mon format that is more readily consumable than the disparate data would have been ( status reporting ( ), also referred to as red, amber, and green (RAG) reporting, has emerged as a common element

is likely to go out of tolerance soon (or is improving from red toward green), and green indicates that the status item is on

Little appears to have been written about RAG status report- ing in the academic literature, yet it is a very common project status reporting approach due to the simple visual interpretation of complex project information and the ability for the observer to quickly discern areas where attention is needed, and where

status report is a dashboard that centralizes the information of many concurrent projects and captures status over time, mak-

Project status reporting has some pitfalls, due mainly to the

( ; ; Smith ; ;

that the status report is subject to a degree of creator and reader bias, where a more negative or positive view of the status of a project may be portrayed by the writer or perceived by the reader ( ; ;

; reporting may be inadequate to fully describe the state of proj- ect progress and health in an agile project setting ( Notwithstanding these potential pitfalls, project status is reported on nearly all projects in one way or another, providing a vast pool of time- phased project status data that may be used to identify early warning signs of project trouble (Haji- Kazemi

Project Portfolio Governance and Monitoring Teller

), bringing about the need to monitor the health of multiple imultaneous projects as a cohort, often referred to as a portfolio (

Hopmere et al. 659

activities employed to manage a collection of projects and pro- grams to achieve strategic business objectives (Petit & Hobbs,

;

to keep the portfolio as healthy as possible, seeking to avoid expensive and damaging project failures ( ; PMI,

few projects of their own and with intimate knowledge of how things have progressed to- date may use expert knowledge and trend analysis ( ) to assess project health and take action to keep a project on track, a portfolio manager generally

projects and may have tens to hundreds of projects to monitor

The need for portfolio managers and other stakeholders to be able to view an entire portfolio has, at least in part, made common the practice of standardizing project status reporting formats and tools that simplify the process of adding all these projects and details of their health into a portfolio dashboard ( dashboard centralizes project status reporting information and assists portfolio managers as they assess projects that may need attention ( ), yet, even with a dashboard, many portfolios

In a review work in the area of early warning signs in projects,

sources ranged across hard and soft, qualitative and quantitative data sources and outlined the various forms of analysis required to

-

sources, such as gut feeling, brainstorming from team insight, and extrapolation from previous projects - tify sources including risk analysis, earned value management, and performance measurement

- ences between projects might mean that there is no single source for identifying early warning signs that would apply to all projects, and that a combination of sources and approaches

carefully applied, performance measurement promise and that it may be possible to identify early signs of possible project challenges through the analysis of “suitable project characteristics” ( - cle, we have designated the RAG project status report elements

Several approaches could be taken to make it more feasible to

more portfolio managers and have them monitor smaller segments of the portfolio, with the hope that this would correctly identify

to simply assign more people to the task of portfolio monitoring due

some portfolio managers may be more experienced than others at identifying and intervening in potentially troubled projects, which

Third, even though project dashboards provide a snapshot view of project health at a given time, they put the onus on the portfolio manager to remember how each of their multiple proj- ects was performing in the past, to compare current status to how they were performing in the last report, and then to form a judg-

the expectation that a portfolio manager could do this for more

Forecasting and Predictive Analytics The use of statistical techniques to predict what may occur in

- ysis of purchasing behavior, customer churn, and predictive maintenance ( ; ;

; ;

( ), disputes ( ), escala- tions ( ), failure ( ), and dura- tion (

Project status reports are created in time order and generally at equal intervals, making them candidates for time series fore- casting analysis ( used to forecast future values from current and past values ( ) and has been widely applied over many years

- ences, social science research, and economics ( ;

; has been used in areas as diverse as economic and business planning, production planning ( ), healthcare ( ), and for such applications as electricity load, sales, interest rates, call volumes, employment levels, patient attendance ( ;

; ; The autoregressive integrative moving average (ARIMA)

approach is one of the earlier families of time series forecasting

exponential smoothing ( is the most general form of time series forecasting technique (

), and does not need expert- level parameter selection to generate useful forecasts ( techniques with varying degrees of complexity and utility have

neural networks, support vector machines, and k- nearest neighbor ( work with ARIMA as we assessed it to be the simplest possible analytical approach for this type of data and the target use of fore-

-

Summary Projects vary greatly in terms of complexity, technology, industry,

Project Management Journal 51(6)660

managers, who may be overwhelmed by the scale and dynamic

projects, is an essential aspect of risk management that aims to limit the negative impact of uncertain events and reduce the prob- ability that these events will occur ( primary focus of risk management and monitoring has been on

- temic (unknown- known and unknown- unknown) risks and has largely failed to take into account concurrent and interrelated risks ( projects to identify risk trigger events and early warning signs (

There is a considerable body of research on causes of project failure and early warning signs, but in most cases, research data has been collected after the projects are completed, from a small set of very large case study projects, or from a small sample group

- viously documented early warning signs approaches have various limitations and suggest little to help with monitoring and predict-

Klakegg

memory and hindsight bias ( ), which described as the tendency to view outcomes in

It also generally ignores the information that was generated during the course of the project that may have given even earlier

project management teams and more stringent governance, they make up only a small sample of the total number of projects

- tantly, we consider that retrospectively collected warning signs like those mentioned above provide limited predictive value

Research Description To address the challenges presented in monitoring large portfolios of projects and drawing upon the work of and the risk management and early warning signs literature, the question arises as to whether a before the fact forecasting system can iden- tify emerging project issues early, allowing an organization to

- tral tool in project health and performance measurement and monitoring and are used to communicate status and progress to key stakeholders, providing a quick view of progress, plans, risks,

- erature was found that reported on the use of project status reports

are created in time order, making them candidates for time series

practice, and building on research done to date in this area, we

Is it possible to forecast project trouble over an actionable fu- ture horizon using project status reports?

This research question encompasses several quite complex forecasting (treatment of data, assessment, and applica-

tion of alternative methods), project trouble (the existence of operational issues, risks, and many other issues experienced within the delivery life cycle), actionable horizon (results need

of status reports as the source of project health information,

Research Design

monitoring large project portfolios is questionable when we consider how poor people generally are at accurately forecast- ing future outcomes, especially where those predictions may be things we do not want or expect to occur ( ;

; ; been shown that expert predictions are often inferior to those generated by simple statistical models ( ;

portfolio would be to provide portfolio managers with a system that combines human judgment and statistical inference, lever- aging the wealth of information collected for each project in project status reports to proactively monitor and forecast future project health and draw attention to potentially troubled proj-

across the entire portfolio, ensuring that all projects are assessed objectively and would be able to constantly monitor all projects in the portfolio and short list projects that need a portfolio man-

Research Data In this research we use data from projects from a single indus- try and technology area and that can be collected and analyzed

lists the perceived issues of previous early warning signs research and shows how we have

industry and project type to be an advantage in that it allows us to avoid some of the issues of past research in similar areas and

Using the terminology for early warning sources, this research focuses on hard, quantitative project performance measurement data documented in project

Hopmere et al. 661

assessments in the form of project status reports collected from three IT portfolio dashboards, each of which tracked separate

study organization is a large multinational company, with sig- -

shows a sample of the collected status report data for a

-

describes the data, labels applied, and forecasting -

Research Approach As mentioned in the literature review, project status reports are created in time order, making them candidates for time series

forecasting analysis ( as we assessed it to be the simplest possible analytical approach

-

To prepare the status report data elements for ARIMA analy- sis, we converted the RAG status entries to integers where green

, the project status reports contain multiple key performance indicators (KPIs) that together provide a view of the health, progress, and status of a

- vidual KPI, based on early experimentation, and considering the systemic interrelatedness of these KPIs and the argument that project behavior could be better explained through this interrelat- edness as a whole rather than through any single KPI (

converted RAG values) for individual status weeks for all proj- ects to create an aggregate score that summarized the health of a

The main concern arising from forecasting each individual

KPIs was forecast individually, the portfolio manager would

each project within their portfolio of often tens to hundreds of

Table 2. Sample of Collected Project Status Data

Project Code Report End Date Overall Scope Schedule Budget Adoption Deliverables People Issues Risks Quality

4136 1/8/2011 A G A G G G G G A G 4136 1/15/2011 A G A G G G G G A G 4136 1/22/2011 A G A G G G G A A G 4136 1/29/2011 A G A G G G G A A G 4136 2/5/2011 A G A G G G G A A A 4136 2/12/2011 R G R G G R G R R A 4136 2/19/2011 R G R G G R G R R A 4136 2/26/2011 R A R R G A A G A G : : : : : : : : : : : : --- --- --- --- --- --- --- --- --- --- --- ---

Key: G = green. A = amber. R = red.

Table 1. Issues of Previous Early Warning Signs Research and Detection Approaches

Issue Issue Description Mitigation in This Research

Post- Facto Research data collected after projects have completed Used data that can be collected live during the course of project delivery

Very Small Samples Data collected from a small number of projects Used data from thousands of projects in very large portfolios

Very Large Projects Data collected from (often) very large projects Used data from projects of various sizes, durations, and complexities

Mixed Industry Data collected from multiple industries and combined in research results

Used data from a single industry area

Mixed Technology Areas Studied projects delivering multiple different technologies

Used data from a single technology area

Project Management Journal 51(6)662

which these KPIs interact with each other (for example, the interaction of scope, schedule, and cost) means that an aggre- gate health score as described above may better describe the

As shown in

For simplicity in this work, we have equally weighted all of the

value is shown as follows and the elements of the formula are described in

calculated as the total of all available status KPIs (ST) for all

scores is maintained and converted to time series in the ARIMA

Status Sum Formula. For Project P and time TS, cal- ...

∑

- age ( ) for its automatic parameter setting

then the accuracy of these forecasts was tested to see how often

shows an example of project status plotted over time

For the purposes of the analytical aspects of this work,

The six- reporting- period horizon forecast values (marked as

- ening) status trends that were of interest when forecasting

approach to performance measurement in predicting early warning signs in the project management con-

value above the alarm threshold turned out to be correct, a true

Table 3. Status Report Entries From Each Data Source

KPI Source 1 Source 2 Source 3

Scope X Schedule X X X Budget X X X Risk X X X Issues X Customer Satisfaction X X Contractor Satisfaction X X Team Satisfaction/People X X X Contract Compliance X X X Process Quality X X Solution Quality X Technical/Deliverables X Adoption X Overall X X X Trend X Total 11 9 10

Table 4. Data Summary and Units of Analysis

Symbol Detail Range – Source 1 Range – Source 2 Range – Source 3 Total/

Summary

S1,2 Data sources S1 S2 S3 P1…NP Projects 1...NP P1 – P7583 P1 – P1639 P1 – P7646 NST Number of Status KPIs in dataset 11 KPIs 9 KPIs 10 KPIs ST1…NST KPI status 1...NST for Week 1...NTS for

P1…NST

ST1 – ST11 ST1 – ST9 ST1 – ST10

TS1…NTS Week 1...NTS for P1…NP TS1 – TS171 TS1 – TS177 TS1 – TS74 SSTS1…NTS Calculated status sum for Week 1...NTS

for P1…NST

0-22 0-18 0-20

Alm1,2 Alarm threshold values Alm1 = SSAllAmber

Alm2 = SSAllAmber + (SSAllRed – SSAllAmber)/2

Alm1 = 11 Alm2 = 16

Alm1 = 9 Alm2 = 14

Alm1 = 10 Alm2 = 15

Int/Ext Project Recipient Internal Customers External Customers External Customers NP Number of unique projects 7,583 projects 1,639 projects 7,646 projects 16,868 projects NTS Number of weeks in unique project 1–171 weeks 1–177 weeks 1–74 weeks Max 177 weeks SR1…NSR Status report entries 81,765 23,568 70,443 175,775 Duration Calendar year coverage of data 8 years

2008–2015 6 years

2010–2016 1.5 years

2016– Max 8 years

Hopmere et al. 663

Research Results

- sitivity or true positive rate (proportion of positives correctly

we found that this approach provided encouragingly accurate,

Discussion Identifying projects that may become troubled in large portfo-

- erenced literature ( ; ;

; Haji- Kazemi & Andersen, ; , , Haji- Kazemi,

,

Figure 1. Project status actuals and forecast plot.

Table 5. Forecast Results at Alarm Threshold 1—Upper 80% Confidence Range Forecasts

Source 1 Source 2 Source 3

Upper 80 Upper 80 Upper 80

Forecast Count Forecast %

Project Count Project %

Forecast Count

Forecast %

Project Count Project %

Forecast Count

Forecast %

Project Count Project %

Projects - - 7,582 - - - 1,639 - - - 7646 - True Positive (TP) 4,522 0.8% 170 2.2% 2,464 1.5% 58 3.5% 8,161 2.7% 163 2.1% True Negative 550,788 96.2% 7,565 99.8% 155,981 94.5% 1,625 99.1% 278,756 93.8% 7,614 99.6% False Positive (FP) 16,604 2.9% 639 8.4% 6,468 3.9% 168 10.2% 8,621 2.9% 346 4.5% False Negative 434 0.08% 57 0.8% 63 0.04% 8 0.48% 1,422 0.47% 210 2.7% TP +FP 21,126 0.04% 673 8.9% 8,932 0.05% 186 11.3% 16,782 0.06% 384 5.6% Sensitivity 0.91 0.97 0.85 Specificity 0.97 0.96 0.97 Accuracy 0.97 0.96 0.96

Project Management Journal 51(6)664

; ; ; Klakegg ; ) have various limitations and

suggest little to help with monitoring projects that are still in

If we compare the forecasting system described here to having portfolio managers manually check all projects in the portfolio

compares several alternative status monitoring approaches that

every week ensures that all projects that will trend into trouble in

Even though all the projects that could trend into trouble would be captured in this large net, portfolio managers monitoring these projects would need to be able to correctly assess all available current and past data and mentally infer whether a project might

scale of such an undertaking for a large project portfolio and the challenge becomes much greater and accurate and actionable

thresholds were quite rare, making this comprehensive checking

checked would not have trended into trouble, whether they were

Rather than attempting to check all projects in the portfolio every week, another approach may be to check a smaller subset

this would greatly reduce the probability of selecting all the potentially troubled projects in each pick and it would take a full

to recall past performance and accurately infer future perfor-

In contrast, the forecasting approach described here avoids the issues of the manual approaches described earlier by applying a single consistent method across the entire portfolio, ensuring that all projects are assessed equally, by constantly assessing all proj- ects in the portfolio, short listing those that may need a portfolio

shows that applying this forecasting approach, even when all true positive and false positive indications are investi-

Results of this research suggest a positive response to the research question, demonstrating that it is possible to forecast project trouble over an actionable future horizon using project

Table 6. Forecast Results at Alarm Threshold 1—Upper 95% Confidence Range Forecasts

Source 1 Source 2 Source 3

Upper 95 Upper 95 Upper 95

Forecast Count

Forecast %

Project Count Project %

Forecast Count

Forecast %

Project Count Project %

Forecast Count

Forecast %

Project Count Project %

Projects - - 7,582 - - - 1,639 - - - 7,646 - True Positive (TP) 4,746 0.8% 167 2.2% 2,499 1.5% 60 3.6% 8513 2.8% 166 2.2% True Negative 533,204 93.2% 6,889 90.9% 149,016 90.3% 1,625 99.1% 273,158 91.9% 7,614 99.6% False Positive (FP) 34,188 6.0% 876 11.6% 13,433 8.1% 273 16.6% 14,219 4.7% 485 6.3% False Negative 210 0.04% 20 0.3% 28 0.02% 3 0.18% 1,070 0.36% 210 2.7% TP +FP 38,934 0.07% 898 11.8% 15,932 0.10% 291 17.8% 22,732 0.07% 524 7.6% Sensitivity 0.95 0.98 0.88 Specificity 0.93 0.91 0.95 Accuracy 0.93 0.91 0.94

Table 7. Comparison of Status Monitoring Approaches. (Source 1 Using Upper 80% Confidence Range)

Approach Minimum Number

of Projects Reviewed

Probability of Choosing All Current/Future Trouble Projects

Probability of Correctly Forecasting

Future Status Effort and

Cost Range

Routinely check all projects 7,582 100% Low High Sequentially check 25% of projects 1,895 25% Low Med/High Check all projects flagged by forecasting

system 673 90% High Low/Med

Hopmere et al. 665

Limitations listed the perceived issues with previous early warning

signs research and showed how we avoided these issues in this

managed by and reported on by people who are open to many biases, and these may be played out in the reported status of some

of a project, very few systems or approaches, whether human or

as automating the status calculations for some KPIs, for example schedule, risk, or budget, which lend themselves to such a quanti-

reported, it would limit the ability of individual people to misre- -

Another potential weakness is that we have chosen to use all project records from the three separate sources even though

organization, process, duration, complexity, scope, technology

projects within each of the three data sets is accepted in this -

we analyzed and reported the results for each data set

From a technical perspective, one limitation of this approach is that each project is forecast entirely independently of all other projects, limiting the inference and application of learn-

does not consider other information that may be known about a

our focus on the dynamic indicators of project health, the largely static factors, such as geography and the technology being delivered, were omitted as they are unlikely to change once set at the start of the project and therefore were deemed to

Last, the potential limitation imposed by the data used in this work all coming from IT industry projects must be

RAG KPI reporting across industries, it seems intuitive that

However, without further experimentation with data from other

Notwithstanding the potential issues discussed here, we found that time series forecasting using IT project status reports

three data sets analyzed, our approach correctly forecasted when future project status would exceed the alarm threshold

time (see Sensitivity rows in

occur quite rarely (see False Positive rows in ), false positives may cause alarm fatigue where project and port- folio managers are subjected to a degree of sensory overload from too many alerts ( ; ;

had this system been operating over the life of the project port- folio, these false positive alerts would have been spread over multiple years, making it unlikely that this relatively small

Another area of potential concern is the occurrence of false negatives, where a project exceeded the alarm threshold, but

Negative rows in ), making up a low percentage -

projects should not be underestimated, however this condition

Conclusion -

managers, who may be overwhelmed by the scale and dynamic nature of the portfolio, presenting an opportunity for a forecast- ing approach that can assist portfolio managers by accurately identifying projects from within the entire portfolio that may

converting them to time series, and then forecasting over a six- reporting- period future horizon presents encouragingly accu- rate and actionable results, correctly forecasting future project

the foundation for an “expert system” to assist portfolio manag- -

tion to potentially troubled projects, which, if applied to the management of live project portfolios, may help portfolio man- agers focus their expert skills on assessing the deeper issues of a project and deciding whether and how to intervene rather than inspecting projects that are not, and probably never will, get

Notwithstanding these encouraging results, simply present- ing accurate forecasts may not be enough to ensure that these

- selves a complex area ( ;

raised, people will act to do something about it ( ), but this is much less common than would be

Project Management Journal 51(6)666

found that many warning recipients considered “dedicated warning products” like the system that could be derived from this research to be irrelevant

“gut feelings” over future- looking external warning systems despite the research that their predictions, even if they are

simple statistical models ( ;

alerts created by this system are indeed used by portfolio

This work contributes to the risk management and early warning signs literature by demonstrating that the statistical analysis of historical IT project status reports can be used to generate useful forecasts of emergent risks, even when these

implying that this approach may be applied in other organiza- -

ect health monitoring within the area of project portfolio management toward “proactive monitoring” by presenting an

more focus on the generation of warnings during the project delivery phase, enabling continuous portfolio supervision, which is more practical and precise than a system reliant on

complexity by applying an analytical approach to project man- agement data that is akin to those that might be applied in some branches of complex systems research, thereby contributing in a limited way to the growing bridge between the project man-

This work opens research avenues into other industry and project types, the application of more complex analytical tech-

such as on- time or on- budget completion and customer satis-

trouble and that were intercepted by the portfolio management team still ended up in trouble? How many projects were proac-

portfolios that have this system in place have higher customer

or manage their portfolios?

Declaration of Conflicting Interests

Funding

References

IFAC Proceedings Volumes, 46

California Management Review, 18 Strategic Manage-

ment Journal, 1

Time series analysis: Forecasting and control

- -

International Journal of Disaster Risk Reduction, 9,

International Journal of Project Manage- ment, 33

- IEEE

Transactions on Engineering Management, 60 -

Journal of Civil Engineering and Management, 19

- AORN Journal, 103

- Science, 243

- International Journal of Forecasting, 22

on time series forecasting techniques for building energy con- Renewable and Sustainable Energy Reviews, 74,

behavioral competencies of information technology project manag- Project Management Journal, 46

- odologies for estimating the likelihood of program implementa-

American Journal of Evaluation, 35

European Journal of Social Psychol- ogy, 5

- IEEE Systems Jour-

nal, 12

- Sciento-

metrics, 95 Overcoming ICT project fail-

ures—A practical perspective

Hopmere et al. 667

- Organizational Behavior and Human Deci-

sion Processes, 110 -

Organizational Behav- ior and Human Performance, 13

Transportation Planning and Tech- nology, 30

International Journal of Project Management, 31

Science, 275

International Journal of Project Management, 36

- International Journal of Operations & Pro-

duction Management, 31 Project management: The manage-

rial process

International Journal of Managing Projects in Business, 6

Interna- tional Journal of Managing Projects in Business, 7

International Journal of Project Management, 33

of early warning signs in front- end stage of projects, an aid to Selected

papers from the 26th IPMA

Project Management Journal, 44

Advances in production management systems: Innovative production management towards sustainable growth APMS 2015. IFIP Advances in Infor- mation and Communication Technology

Canadian Journal of Civil Engineering, 32

Psychological Review, 118

European Journal of Operational Research, 253

- book

Journal of Enterprise Information Management, 30

Forecast: Forecasting functions for time series and linear models -

-

MIS Quarterly, 33 -

agement involvement impacts project portfolio management per- International Journal of Project Management, 28

Information Sys- tems Management, 23

MIS Quarterly, 24

Journal of the Association for Information Systems, 15

MIT Sloan Management Review, 55

Advanced project management: Best practices on implementation

- -

International Journal of Project Management, 30

A contingency- based view of

- Early warning signs in complex projects -

IEEE Transactions on Visualiza- tion and Computer Graphics, 17

International Statisti- cal Review/Revue Internationale de Statistique, 44

- ing practices in the preacquisition phase for federal intelligence

Project Management Journal, 39

- , 9

Project Management Journal 51(6)668

Early warnings: A phenomenon in project man- agement

International Journal of Project Management, 15

International Journal of Project Management, 19

Project portfolios in dynamic environ- ments: Organizing for uncertainty

IEEE Transactions on Engineering Management, 37

A guide to the project management body of knowledge (PMBOK® guide) – Third edi-

Pulse of the profession®: The high cost of low performance

Agile practice guide

A guide to the project management body of knowledge (PMBOK® guide) – Sixth edi-

The standard for port- folio management –

Pulse of the profession®: The project manager of the future—Developing digital- age pro- ject management skills to thrive in disruptive times

Pulse of the profes- sion®: The future of work—Leading the way with PMTQ

The standard for risk management in portfolios, programs, and projects

Current Opinion in Anaesthesiology, 28

Inter- national Journal of Managing Projects in Business, 2

Communications of the ACM, 50

Journal of Business Logistics, 36

Predictive analytics: The power to predict who will click, buy, lie or die

Information Sys- tems Journal, 13

Journal of Management Information Systems, 18

Information & Management, 44

-

7th International Conference on Knowledge Management in Organizations: Service and Cloud Computing

- Psychological Science in

the Public Interest, 1

- Project Management

Journal, 45 Project

Management Journal, 44 The handbook of project based management

Hidden brain

Euro- pean Journal of Operational Research, 259

Journal of Personality and Social Psychology, 39 Post- project reviews to gain effective lessons

learned Project

Management Journal, 48 -

Project Management Journal, 43

Author Biographies

Michael Hopmere

aspects of project management, with particular focus on advanced analytics techniques that may assist project and port- folio management professionals in their daily work managing

Professor Lynn Crawford Management Program in the School of Project Management at

University, School of Management (UK) and Honorary Adjunct

researching projects and working with leading corporations and government agencies, assisting them in developing organi- zational project and program management capability, facilitat- ing design thinking approaches and the sharing of knowledge and best practices through global project communities and

Fellow of IPMA and APM, and a recipient of the Sir Monty

Hopmere et al. 669

Finniston Award for lifetime contribution to project manage-

Crawford has been actively engaged in the development of

and was instrumental in the formation of the Global Alliance

Dr. Michael S. Harré is Senior Lecturer in the Complex Systems Research Group in the School of Civil Engineering at

focusing on the psychology of games and the tipping points

- cial intelligence that could simulate the human- like psychology

-

grant that will simulate the psychology of buying and selling in housing markets in order to understand under what circum-

Copyright of Project Management Journal is the property of Sage Publications Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.