Help & reference for software reliability prediction
Requs AI Predict turns a software size and development-practice survey into a full reliability growth picture — defect density, defect counts, failure rate, MTTF, availability and reliability — using a machine-learning model trained on historical projects.
Every tab in Requs AI Predict
The application is organized around a project file. Open or create one, then work through these tabs in order.
Dashboard
See the results. Visible once a project is open and the required inputs are confirmed.
Software FMEA tab
The software FMEA module — visible only when purchased and a project is open.
Estimation
Reliability growth module — trends actual test data against Exponential and Logarithmic models to forecast MTBF.
The 7 software reliability prediction formulas
Each stage below feeds the next — size scales the defect count, the defect profile drives the failure rate, and failure rate resolves into availability and reliability.
Predict size
Estimating software size using detailed source-lines-of-code estimates, or via people-hours, normalized to one base language.
- Normalized effective KSLOC
- Upper / lower size bounds
Predict defect density
Completing the development-practice survey so the machine learning model can predict defect density from your assessment.
- Predicted defects per KSLOC
- Sensitivity analysis inputs
Predict defect counts
Total predicted defects = normalized effective size × predicted defect density.
- Total predicted defect count
- Upper / lower defect bounds
Predict defect profile
Predicting when defects will be detected for the first time, across the Rayleigh growth curve.
- Monthly discovery profile
- Failure rate calculation
Predict failure rate / MTTF
Deriving failure rate, MTTF, critical failure rate and MTTCF from the defect profile, duty cycle and installed sites.
- MTTF / MTTCF
- Availability calculation
Predict availability
Predicting availability from MTTCF and restore time (MTSWR) — the measure that matters for continually-operating systems.
- Monthly availability growth
- Reliability calculation
Predict reliability
Predicting the probability of failure or success over a specific mission time.
- Reliability growth summary
- Export to the Dashboard
Nine steps from LRU to release decision
Work through these in order the first time you set up a project.
Identify the software LRUs
Identify the software LRUs (Line Replaceable Units) in your system. See IEEE 1633 for more guidance.
Create a project per LRU
Press “New” on the Project tab for each LRU. Different LRUs can have different duty cycles, development organizations or languages — don’t model them as one project.
Confirm the required inputs
Go to Survey → Required Inputs. Every input must be edited or confirmed — size estimates have no default and must be entered. Ask software engineering if you don’t know an answer.
Answer every category
Visit each remaining Survey tab. Answering some questions in every category beats answering all questions in just one. Leave a response as “U” if it’s unknown.
Run the Dashboard
Press “Refresh all inputs” on the Dashboard tab. The frames show the average of each estimate over the life of this software version, up to the next feature release.
Read the growth profiles
View the remaining-fault and remaining-critical-fault profiles, plus failure rate, MTTF, MTTCF, reliability and availability. Save any graph with the disk icon.
Pull the growth summary
Press “Reliability Growth Summary” for every metric in tabular form, and export it if you need it elsewhere.
Export the project
Back on the Project tab, press Export to export all inputs and the average predictions shown on the Dashboard.
Revisit as things change
Revisit your inputs whenever size estimates, required inputs or development practices change, or you learn more. More accurate answers make for a more accurate prediction.
Find the formula, tab, or step you're working with.
Jump straight into the prediction pipeline, or start from the quickstart guide.