Menu

Glossary

Predictive Analytics

Predictive analytics is the use of historical data, statistical models, and machine learning to forecast future outcomes such as rent, occupancy, or default risk. It answers "what will happen next" rather than "what happened," turning a portfolio's past leasing, market, and operating data into forward estimates that guide underwriting and asset management decisions.

How Predictive Analytics Works

Predictive analytics works by training a model on historical patterns, then applying it to current inputs to project a future value. Per Gartner, the defining trait is a focus on prediction rather than description. A model learns the relationship between drivers, such as absorption, employment, and lease expirations, and an outcome, such as next-year rent, then extrapolates.

The workflow has three stages: assemble clean historical data, fit a model that minimizes error against known outcomes, and score new inputs to generate a forecast with a confidence interval. In commercial real estate the inputs span lease data, submarket supply, demographic trends, and macro indicators. The output is a probability or a range, not a certainty, which is why teams pair every forecast with the assumptions behind it.

Stage

Input

Output

Data assembly

Lease history, market comps, demographics

Clean training set

Model fitting

Known past outcomes

Weighted drivers, error estimate

Scoring

Current property and market inputs

Forecast with confidence range

Why Predictive Analytics Matters

Predictive analytics matters because it moves a team from reacting to reporting toward acting on a forecast. Per a 2025 Deloitte outlook, 81% of CRE firms named data and technology a top spending priority, and predictive tooling is the layer that turns that data into a decision. The value is earlier, better-informed choices on rent, hold, and capital.

The gains are real but must be framed honestly. Per RTS Labs, predictive analytics can increase investment decision accuracy in a reported 15% to 25% range and surface opportunities faster than manual review. The risk is overconfidence: a model trained on a stable market fails when the regime shifts, and a forecast stated without its confidence interval reads as fact when it is an estimate. A prediction is only as good as the data and the assumptions feeding it.

Example

Consider a submarket rent forecast. A model is trained on three years of quarterly data linking net absorption to rent growth, and it estimates that each 1% of positive net absorption corresponds to 0.8% of rent growth over the next year. The analyst applies current conditions.

Step

Input

Calculation

Result

Current in-place rent

$30.00 per sq ft

baseline

$30.00

Projected net absorption

3.0%

model input

3.0%

Model coefficient

0.8% rent growth per 1% absorption

3.0% x 0.8

2.4%

Forecast rent

$30.00 x (1 + 0.024)

$30.00 x 1.024

$30.72

The model forecasts $30.72 per square foot next year, a 2.4% rise driven by projected absorption. A confidence interval of plus or minus 1.5% puts the range at roughly $30.26 to $31.18. The analyst underwrites to the low end and treats the point estimate as the base case, not the plan. The coefficient here is illustrative; a production model would be fit to the specific submarket's actual history.

Variations and Edge Cases

Predictive analytics spans several CRE use cases, each with a different target and data source. The table below maps the common applications.

Application

Target forecast

Primary data

Rent and occupancy

Next-period rent, lease-up pace

Lease history, absorption, supply

Predictive maintenance

Equipment failure timing

IoT sensor readings

Tenant default risk

Probability of delinquency

Payment history, credit signals

Acquisition targeting

Likelihood a property sells

Owner behavior, market conditions

Predictive Analytics vs Descriptive Analytics

Predictive analytics is often confused with descriptive analytics, but they answer different questions. Descriptive analytics answers "what happened," summarizing past and current data into dashboards and reports. Predictive analytics answers "what will happen next," using statistical models and machine learning to forecast outcomes the historical record does not yet contain.

The distinction is direction in time. Per Gartner, business intelligence and its descriptive methods evaluate the current state, while predictive analytics extends beyond it to future events. A rent roll summary is descriptive; a forecast of next year's rent from that rent roll plus market data is predictive. Most CRE teams need both: description to know where they stand, prediction to decide where to move.

Frequently Asked Questions

What is predictive analytics in commercial real estate?Predictive analytics in commercial real estate is the use of historical data and machine learning to forecast outcomes such as rent, occupancy, tenant default, or maintenance needs. It answers "what will happen next," turning past leasing and market data into forward estimates that guide underwriting and asset management.

What can predictive analytics forecast in CRE?Predictive analytics commonly forecasts next-period rent and occupancy, tenant default probability, equipment failure timing for maintenance, and the likelihood a property will sell. Per RTS Labs, it can increase investment decision accuracy in a reported 15% to 25% range when the underlying data is clean.

How is predictive analytics different from descriptive analytics?Descriptive analytics answers "what happened" by summarizing past data into reports and dashboards, while predictive analytics answers "what will happen next" using statistical models. The difference is direction in time: one reports the record, the other forecasts beyond it.

Related Terms

  • Machine Learning Model

  • Absorption Rate

  • Vacancy Rate

  • Underwriting Model

  • Submarket