Selected project

Jakarta air quality simulation from real AQI data.

Nirkabel turns Jakarta AQI records into a pollution prediction and simulation workflow: collect the dataset, model pollutant conditions, simulate changing air quality, then validate the output against ground-truth readings.

Role: Full-stack builder
Stack: Data simulation, dashboard UI
Contribution: Inputs, metrics, workflow, validation
Outcome: 0.8621 overall score

Collected Jakarta AQI data

Built the project around a large AQI dataset from Jakarta so the simulation reflected real locations, time windows, and pollutant behavior instead of demo-only values.

Modeled pollution states

Designed an input flow for location, time of day, pollutant type, and pollutant value, then converted those signals into AQI labels such as Sedang and Tidak Sehat.

Simulated and visualized changes

Added 24-hour AQI trend views and step-based simulation logic so the project could show how readings move over time and where unhealthy conditions appear.

Validated against ground truth

Compared simulation output with real AQI readings and summarized model quality through accuracy, precision, recall, F1 score, and agreement bands.

Interface evidence

Readable screens from the simulation workflow.

The screenshots are presented at full page width so input fields, charts, validation metrics, and dashboard comparisons can be read clearly.

Nirkabel pollution prediction input interface — Prediction interface for location, time, pollutant, and value inputs. Open full-size

Nirkabel AQI trend chart showing unhealthy pollution level — 24-hour AQI trend with selected time and health recommendation. Open full-size

Nirkabel model performance table — Model result summary: accuracy, precision, recall, F1 score, loss, and overall score. Open full-size

Nirkabel triangulation simulation workflow diagram — Simulation workflow for updating points, running triangulation, and stepping through results. Open full-size

Nirkabel comparison dashboard between ground truth and simulation — Dashboard comparing ground truth data with the OMNeT++ simulation output. Open full-size

Prediction flow

From model input to AQI result.

Load modelPrepare the trained prediction model.
Collect inputsRead location, time, pollutant, and value.
Predict categoryConvert the pollutant signal into an AQI label.
Validate outputCompare the result against expected AQI behavior.
Visualize resultShow the predicted category and trend context.