Calibration is a vital process in Fligence Planning to refine model accuracy, involving adjustments of multiple parameters against actual data benchmarks. Because the QSI calibration is intricate and time-consuming, we provide some reports and an automatic share gap calibration to expedite the procedure.
Figure 1 shows the calibration interface and menu. The system simplifies calibration with a one-click feature on the Advanced Setting tab of the Scenarios panel, which can be triggered from the menu Planing > Scenario Analysis. However, preparatory and verification steps are essential.
Benchmark Data
Actual traffic and fare data by airline and service type are needed for benchmarking. This can be generated in two ways:
- Your own Market Size data: Import from a CSV file at the Actual Traffic & Fare interface shown in Figure 2, which can be triggered from the Planning > Calibration Related > Actual Traffic & Fare. When importing actual data, it must be formatted to match the system’s required structure. The table for import should include:
- Mkt_AL: The IATA code for the marketing airline.
- Org: The IATA code for the origin airport.
- Dst: The IATA code for the destination airport.
- Traffic_Type: Categorized as ‘Local’ or ‘Connect’ for non-stop and connecting services.
- Pax: The passenger count for the specified forecast period.
- Fare: The average fare per passenger in USD for the forecast period.
- Our estimated Market Size data: Select from the system’s internal data using the calendar control next to the Calibrate button in Figure 1 for data from January 2015 up to six months ahead. After the calibration, the internal benchmarking data is also shown in the Actual Traffic & Fare table.
Data in the Actual Traffic & Fare table is always used for calibration. If that table is not empty and you want to use our estimated Market Size data, you must clear it first. This is because if you previously loaded your own Market Size data, the system will not overwrite it with our estimates.
To use our estimated market size data in the absence of your own, click Calibrate in the Advanced Setting tab within the scenario analysis panel. This action will populate the Actual Traffic & Fare table.
Initial Scenario and Forecast Deviations
To initiate the Airline Deviation and Market Deviation reports, navigate to Planning > Calibration Related and select either Airline Deviation or Market Deviation, as depicted in Figure 1. There reports require actual market size data; if they appear empty, verify that the Actual Traffic & Fare table is populated. Instructions for populating this table with actual market size data are detailed in the preceding Benchmark Data section.
Run a scenario without any Airline Adjustments first to identify deviations, as shown in Figure 3 and Figure 4.
The Forecast Deviation by Marketing Airline report, accessible via Planning > Calibrated Related > Airline Deviation, provides insights into discrepancies between forecasted and actual figures, as illustrated in Figure 3. The report includes:
- Dom_Mkt_AL: IATA code for the dominant marketing airline.
- Pax: Passenger count from the current scenario for this airline.
- Pax_Benchmark: Passenger count from the actual data for comparison.
- Abs_Pax_Diff_Percent: Absolute percentage difference in passenger count against the benchmark, aggregating discrepancies across O&D records. This metric is conservative because errors on O&D markets are accumulated using the absolute values.
- .Rev: Revenue from the current scenario for this airline.
- Rev_Benchmark: Revenue derived from the actual data.
- Abs_Rev_Diff_Percent: Absolute percentage difference in revenue against the benchmark, highlighting variances in absolute terms at the O&D level.
The report at Planning > Calibration Related > Market Deviation allows you to scrutinize variances at the Origin & Destination (O&D) level of the host airline. Detailed comparisons of a sample scenario are presented in Figure 4. This table only showcases forecast deviations of the Host airline(s). If the table appears empty, please add one or more Host Airlines on the Scenario Analysis panel, click Save in the QSI Settings tab, and then navigate back to this report.
Calibration
Click the Calibrate button in the Advanced Settings tab to generate a new Airline Adjustment table, as shown in Figure 5.
Verification
Re-run the scenario with the newly generated Airline Adjustment table and compare deviations. Figure 6 and 7 depict exemplary outcomes post-calibration, showcasing reduced discrepancies at both the airline and O&D levels. Initially, deviations hovered between 20-30%; following calibration, these have notably diminished to below 2%, illustrating the effectiveness of the calibration process.