With January half over and the last of the visitor arrival data for 2022 uploaded and discussed, I thought it was time to turn my attention to this year. 2022 showed a significant turnaround in arrivals with the total arrivals at the end of the year being 63% of 2019’s totals (442,576) after dismal arrivals in 2020 and 2021.
The predictive model that I created using Power BI worked very well, with every monthly arrival total falling within the confidence interval of the prediction. However, for 2023 I would like to test the model’s predictive qualities in the following ways;
- Compare the 12-month prediction made by the model at the start of this year to what it predicts when it is updated with the monthly totals (adding the actual arrivals for the month updates the model’s prediction). Will the prediction made at the start of the year still be valid as the year continues?
- Compare the closeness of fit between the 12-month and monthly updated predictions. Which one works better overall? Which is the closer fit to the actual arrivals?
To track this and allow anyone to see a snapshot of the information, I created a simple online dashboard using Google Sheets (Link: Tourist arrivals dashboard 2023 — Google Drive), where you can see the performance of both models as the year progresses.
As one would expect, both the 12-month and month-to-month models will predict the same number of arrivals for January 2023. Both predict that the arrivals for January 2023 will be 53,359, with a 95% confidence interval of 37,880–68,838.
Results for January
Pre-amble notes:
i) The results from the Barbados Statistical Service placed the arrivals in different categories that they have been in over the last 4 years ( 2018–2022). This required a remake of the data model to allow for the new and old categories to exist together. For longer term analysis and storage of the data, this will have to be kept in mind.
12-month prediction results
Monthly updated prediction results
The total number of tourist arrivals for January 2023 was 58,492. This was 5,133 more arrivals than predicted (Prediction: 53,359 The actual arrivals do fall within the 95% confidence window (37,880–68,388). The dashboard has been updated to reflect these results as well as to add the February prediction for the month-to-month update model.
February prediction (Start of 2023 model)
Predicted arrivals: 55,197.
95% confidence interval: 34372–76022
February prediction (Monthly updated model)
Predicted arrivals: 59,817.
95% confidence interval: 44411–75223
Difference between model Predictions for February (Start of 2023 model -Monthly updated model)
Predicted arrivals: -4620