Construction Cost Control Dashboard
Laura Camila Guayacan Ardila | Cost drivers, scenario analysis, estimate range and unit cost variance
Read first: the data comes from a public synthetic dataset (6,123 projects), so figures illustrate the method and are not market benchmarks. Numbers are calculated from the regression equation. The "Interpretation" boxes are text written from the verified results; no AI model runs in this page.
1. Cost drivers
Change in project cost for a typical (+1 standard deviation) increase in each factor. Linear regression, 6,123 projects.
Statistically significant (p < 0.05) Not significant Negative effect
InterpretationArea explains most of the cost: a typical increase in area moves cost by about USD 486k, more than ten times the next factor. Equipment, floors, labor index and material index are significant but small. Weather, risk, complexity and the categorical choices (type, location, construction, foundation) cannot be distinguished from zero in this dataset.
2. Scenario analysis and estimate range
Adjust the project. The baseline is the median project. The estimate uses the log-log model (R² 0.87).
-
Estimated cost (point)
-
Change vs baseline
-
Estimate range (about 80%)
-
Unit cost (USD per sq ft)
Sensitivity: cost change for +10% in each driver
Interpretation
3. Unit cost variance watchlist
Projects whose cost per sq ft is above the IQR screening threshold (USD 194.85 per sq ft): 423 of 6,123 (6.9%). SQL analysis in BigQuery.
Share of projects flagged, by area
Equipment cost as a share of total cost
Highest unit cost projects
| Project | Type | Area (sq ft) | Cost (USD) | USD per sq ft | Equipment (USD) |
|---|
InterpretationThe flag is a size effect, not a project type effect: 84% of projects under 2,000 sq ft are flagged and none above 5,000 sq ft. Equipment cost is a near-constant USD 100k, so it weighs 32% of cost on small projects and 7% on large ones. Removing it halves the flagged projects (423 to 205). The rest is unexplained here; other fixed costs such as mobilization or permits are a plausible, untested hypothesis. For small projects, review fixed costs and allow a larger contingency.