Computer Engineering is coming soon

We're still writing and reviewing the notes for every subject, and we'll open it when every subject is ready. Meanwhile you can practise aptitude, ask Aria about any subject, or see the branches that are ready.

Project planning and size metrics

Project planning uses size metrics (LOC/KLOC, Function Points, Use‑Case Points) and estimation models (COCOMO, productivity rates) to predict effort, schedule and team size for software projects.

Drafted with Aria, reviewed by the AiCanCode.org team · Updated . How we write and check our notes. Spotted an error? Use Give Feedback at the bottom of the page.

Why it matters

Two sentences: Engineers use project planning and size metrics to estimate how much work a software project will take (effort in person‑months), how long it will take (duration) and how many people are needed (staff). Good size metrics and models (COCOMO, Function Points) give data for budgeting, scheduling, risk management and staffing decisions.

Key ideas

Start simple and define terms before use.

  • Project plan: a documented set of decisions about scope, schedule, resources and risks for a software project. It contains estimates for effort, duration and staff.
  • Size metric: a measurable quantity that represents the size of a software product; used as an input to estimation models. Examples: LOC (lines of code), KLOC (thousand LOC), Function Point (FP).
  • LOC (lines of code): count of executable source lines. Define counting rules before use (exclude comments, blank lines or include them consistently).
  • KLOC: 1 KLOC = 1000 LOC. Unit used in many estimation formulas.
  • Function Point (FP): a language‑independent measure of software functionality from the user's perspective. Calculated from counts of inputs/outputs/queries/files and a complexity adjustment.
  • Unadjusted Function Point (UFP): the sum of weighted counts of the five component types (External Inputs, External Outputs, External Inquiries, Internal Logical Files, External Interface Files). You must choose correct complexity weights.
  • Value Adjustment Factor (VAF): calculated from 14 general system characteristics (GSCs) with scores 0–5. The adjusted FP = UFP × VAF.
  • Effort: the total work required, measured in person‑months (one person working one month). Effort is what models predict.
  • Productivity: measured as LOC/person‑month or FP/person‑month; productivity = size / effort.

Models and assumptions

  • COCOMO (Constructive Cost Model): a parametric model that estimates effort from size. The basic form (Basic COCOMO) assumes effort follows a power law of size because of communication and integration overheads as size grows. Basic COCOMO classifies projects into Organic, Semi‑detached and Embedded types; each type has different empirical coefficients. Typical Basic COCOMO coefficients: Organic (a = 2.4, b = 1.05, c = 2.5, d = 0.38); Semi‑detached (a = 3.0, b = 1.12, c = 2.5, d = 0.35); Embedded (a = 3.6, b = 1.20, c = 2.5, d = 0.32). Definitions: Organic = small, simple projects with experienced teams and flexible requirements; Semi‑detached = intermediate projects with mixed experience and moderate constraints; Embedded = large, complex projects with tight hardware, software, and regulatory/operational constraints. Intermediate COCOMO multiplies the basic effort by an Effort Adjustment Factor (EAF), the product of selected cost‑driver ratings (E = a · Size^b · EAF). Detailed COCOMO further decomposes effort across life‑cycle phases. COCOMO II (used today) keeps the form E = A · Size^B · (product of EM_i), with scale factors in B and effort multipliers EM_i; check your syllabus on which version to use in exams.

  • Function Point Analysis (FPA): counts functionality from the user view; assumes a roughly linear relationship between FP and effort, modulated by productivity and environment factors.

  • All models assume the project scope and requirements are reasonably stable or that the estimate explicitly accounts for volatility. If requirements change, re‑estimate.

What to draw in an exam where a diagram is required

  1. Effort vs Size (log‑log): draw X axis labelled "Size (KLOC)" and Y axis labelled "Effort (person‑months)". Plot a curve rising faster than linear (convex upward on linear plot). If using log‑log, draw straight lines for different COCOMO types with different slopes.
  2. FP counting diagram: draw boxes for components (EI, EO, EQ, ILF, EIF) and annotate counts and complexity weights; show sum = UFP, then show VAF calculation and FP = UFP×VAF.

Derivations universities ask for

  • Show derivation of basic COCOMO formula form from regression on historical data: assume E = a·(Size)^b, take natural logs: ln E = ln a + b·ln(Size). Fit ln a and b by linear regression on ln(Size) vs ln(E) from empirical projects.
  • Show how average staffing = Effort / Duration, and that curve for duration is often D = c·E^d in COCOMO where constants come from regression.

Formulas

E = a · (Size)^b — Effort E in person‑months; Size in KLOC; a, b are empirical coefficients; applies to Basic COCOMO.

D = c · (E)^d — Development time D in months; c, d are empirical coefficients; applies in COCOMO for duration estimation.

UFP = Σ (count_i · weight_i) — Unadjusted Function Points; count_i are counts for the five component types; weight_i are standard complexity weights.

VAF = 0.65 + 0.01 · DI — Value Adjustment Factor; DI is the sum of 14 GSC ratings (each 0–5); unitless; applies in IFPUG Function Point calculation. DI ranges 0–70 so VAF is between 0.65 and 1.35.

FP = UFP · VAF — Adjusted Function Points; unit = FP.

Productivity = Size / Effort — Size in LOC or FP; Effort in person‑months; productivity units LOC/person‑month or FP/person‑month.

Staff size ≈ Effort / Duration — persons (average staffing).

Notes: when using LOC/FP conversion you need a language factor (LOC per FP) specific to the implementation language.

Worked examples

Example 1 — Basic COCOMO (numerical) Problem: An organic project size = 32 KLOC. Use Basic COCOMO coefficients for Organic: a = 2.4, b = 1.05, c = 2.5, d = 0.38. Estimate effort, duration and average staff. Steps:

  1. Formula: E = a · Size^b.
  2. Compute Size^b: 32^1.05 = exp(1.05·ln(32)) ≈ exp(1.05·3.4657) ≈ exp(3.6390) ≈ 38.01.
  3. Effort: E = 2.4 · 38.01 ≈ 91.22 person‑months.
  4. Duration: D = c · E^d = 2.5 · (91.22)^0.38. Compute (91.22)^0.38 = exp(0.38·ln(91.22)) ≈ exp(0.38·4.512) ≈ exp(1.7146) ≈ 5.55. So D ≈ 2.5 · 5.55 ≈ 13.88 months.
  5. Average staff: Staff = E / D ≈ 91.22 / 13.88 ≈ 6.57 ≈ 7 persons. Answer: Effort ≈ 91.2 person‑months, Duration ≈ 13.9 months, Average staff ≈ 7 persons.

Example 2 — Function Point counting and LOC conversion Problem: A small system has: 10 External Inputs (average complexity weight 4), 6 External Outputs (avg weight 5), 4 External Inquiries (avg weight 4), 5 Internal Logical Files (avg weight 10), 3 External Interface Files (avg weight 7). GSC sum DI = 30. Use VAF formula and convert to Java using 53 LOC/FP. Steps:

  1. UFP = (10·4) + (6·5) + (4·4) + (5·10) + (3·7) = 40 + 30 + 16 + 50 + 21 = 157.
  2. VAF = 0.65 + 0.01·DI = 0.65 + 0.01·30 = 0.65 + 0.30 = 0.95.
  3. FP = UFP · VAF = 157 · 0.95 = 149.15 FP (do not round yet).
  4. LOC estimate (Java): LOC = FP · 53 = 149.15 · 53 = 7904.95 ≈ 7,905 LOC. Answer: ≈149.15 FP → ≈7,905 LOC (Java).

Example 3 — Using productivity to get effort Problem: If team productivity is 6 FP/person‑month and project size is 180 FP, estimate effort and approximate schedule if team optimum staffing leads to a 6‑month duration. Steps:

  1. Effort = Size / Productivity = 180 / 6 = 30 person‑months.
  2. If we plan duration D = 6 months, average staff = Effort / D = 30 / 6 = 5 persons. Answer: Effort = 30 person‑months, Staff = 5 persons for 6 months.

Common mistakes

  • Forgetting to apply the VAF in Function Point calculations or using wrong DI sum.
  • Using COCOMO coefficients inappropriate for project type (Organic vs Embedded) or forgetting they are empirical.
  • Rounding too early in multi‑step calculations (keep extra precision, round final answer).
  • Confusing person‑months with persons (person‑months / months = persons).
  • Assuming staff and schedule trade off linearly; D = c · E^d and Brooks's Law warns that adding people to a late project can make it later.
  • Using LOC/FP conversion factors from a different source without noting they vary by language and organisation.

For your exam

Typical question types and what full‑marks answers include:

  • Definitions: define LOC, KLOC, FP, UFP, VAF, Effort, Productivity clearly and concisely.
  • Short notes: explain COCOMO assumptions and project classifications; show when FP is preferable to LOC.
  • Derivations: show taking logs of E = a·Size^b and linear regression idea; explain how duration formula arises in COCOMO.
  • Numericals: given size and coefficients, compute effort and duration step by step; show units and rounding.
  • Programs/algorithms: rarely asked; more often you trace FP counting or a use‑case point calculation.
  • Diagrams: draw Effort vs Size and label axes and project type curves; for FP show component boxes and counts.

A full marks numeric answer: list formulas, substitute numbers with units, show intermediate calculations (logs or powers), give final numbers with units and reasonable rounding, and state assumptions (which coefficients, LOC/FP factor, counting rules).

Quick check

  1. What is the formula for adjusted Function Points from UFP and DI?
  2. In Basic COCOMO the effort E is proportional to Size^b. What is the method used to estimate a and b from historical data?
  3. If a project is estimated at 60 person‑months and planned for 10 months, what is average staff?

Answers: 1) FP = UFP · (0.65 + 0.01·DI). 2) Take natural logs ln E = ln a + b·ln(Size) and use linear regression on historical ln(Size), ln(E). 3) 6 persons (60 / 10 = 6).

Finished this topic? Mark it so your progress, study plan and readiness keep up.

Stuck on something here?