Results Analyzer

July 10, 2026 · View on GitHub

The Results Analyzer opens a results database produced by a KSL simulation and helps you explore and compare what's inside: experiments, responses, comparisons, diagnostics, and time series. It's a read-and-report workbench — you don't run models here, you analyze runs you already made.

You will need: Java 21 and a KSL results database (a .db file). You get one by running a model with the database enabled in the Single, Scenario, or Experiment apps. New here? Read Common UI & concepts.

What you'll be able to do

  • Open a results database and browse its experiments and responses.
  • Compare experiments with box plots, confidence intervals, and MCB.
  • Generate within-replication diagnostics, time-series plots, and summaries.
  • Export the data to Excel, CSV, or SQL.

1. At a glance

Results Analyzer — Database tab

Open a database with Open Database…, browse it on the Database tab, then use the analysis tabs to generate reports (which open in your browser).

Use this app when…Use a sibling app when…
You want to analyze runs you already made.You want to run a model → Single / Scenario / Experiment
You want to compare experiments saved in a database.You're comparing scenarios you're running now → Scenario app

2. Before you begin

You need a results database. Any of the run apps can produce one — enable the KSL database option before simulating, and look under your working directory's output/ folder for the .db file. This guide uses a database with two M/M/1 experiments (OneServer, TwoServers), each with 30 replications.

3. A guided tour of the window

Guided tour of the Results window

  1. Menu bar — File (Open Database, Export, Set Working Directory) and Help. A header line summarizes the open database ("MM1Study · SQLite · 2 experiments · 12 responses").
  2. Tabs — Database, Compare Experiments, Within-Replication, Time Series, Histograms & Frequencies, Experiment Summary.
  3. Experiments / Responses tables — on the Database tab, the left table lists experiments; selecting one fills the right table with its responses.
  4. Status bar — messages and the working directory.

4. Tutorial — explore and compare two experiments

Step 1 — Open the database

Click Open Database… (or File → Open Database). Choose the kind — File or Directory (SQLite / Derby)… or Postgres Server… — then browse to the database: a .db file for SQLite, the database folder for Derby, or connection details for Postgres. The header updates to show the database name, type, and how many experiments and responses it holds.

Step 2 — Browse the metadata (Database tab)

Select an experiment on the left; its responses appear on the right with their type (TIME_WEIGHTED, OBSERVATION, COUNTER) and replication count. This tab is a navigator — use the analysis tabs to actually generate reports.

Step 3 — Compare experiments

Results Analyzer — Compare Experiments tab

On Compare Experiments, tick the experiments to include, then Configure… one of:

  • Box Plot — per-replication distributions of a response, side by side.
  • Multiple Comparison Analysis (MCB) — pairwise differences and MCB intervals; the rigorous "which experiment wins?" report (needs ≥ 2 experiments with equal reps).
  • Confidence Intervals Plot — each experiment's mean ± CI for a response.

Each opens a configure dialog (pick the response, confidence level, output location), then Generate writes an HTML report and opens it in your browser.

Reading the results

The comparison report this tab produces is exactly the kind shown in the Scenario guide's MCB example — an alternatives table, a box plot, pairwise differences, and MCB intervals identifying the best experiment with statistical confidence. For our two-server-count experiments, MCB on System Time would identify TwoServers as the lower-cost-in-time configuration.

Other tabs answer other questions: Within-Replication (histograms, Q-Q normality checks for one experiment), Time Series (per- and across-replication plots), Histograms & Frequencies (in-model histogram outputs), and Experiment Summary (cross-replication statistics for one experiment).

5. Reference — every tab explained

TabWhat it's for
DatabaseBrowse experiments and their responses (metadata navigator).
Compare ExperimentsBox plots, MCB, and CI plots across selected experiments.
Within-ReplicationHistograms, observation plots, and Q-Q/P-P normality checks for one experiment.
Time SeriesPer-replication overlays and across-replication summaries.
Histograms & FrequenciesHistogram/frequency responses collected during the run.
Experiment SummaryCross-replication summary statistics for one experiment.

6. Common tasks

TaskHow
Open a different databaseOpen Database… (file/folder, or a Postgres server)
Export the dataFile → Export → Excel, CSV, or SQL
Set the confidence levelIn each analysis's Configure… dialog
Find the database a run producedLook under <working dir>/output/.../*.db

7. Troubleshooting & gotchas

SymptomCauseFix
"not a valid database" on openThe file isn't a KSL SQLite/Derby database, or is still being written.Pick the finished .db file (or Derby folder); ensure the run completed.
MCB option won't generateFewer than two experiments, or unequal replication counts.Include ≥ 2 experiments with the same number of replications.
Responses table is emptyNo experiment is selected.Click an experiment row on the left.
A response has no time seriesIt wasn't collected as a time-series response.Only responses flagged Time Series support those plots.

8. See also