Monthly arrivals forecast 24 months ahead: Holt-Winters on logs, fitted from the post-pandemic reopening, with 80% and 95% ranges that widen with the horizon. Before any forecast is shown, the last 12 published months are forecast blind and the error is published beside it.
Japan
How it is made
backtest_mape_pct7.6fitted_from2022-12horizon_months24made2026-10-20methodHolt-Winters on logs, damped slope 0.9, season started from 2017 to 2019Each row of data, and what meta.method carries beside it, so a number is never without its source.
| Field | Type | Example | What it means |
|---|---|---|---|
period | string | 2026-09 | The month forecast, YYYY-MM. |
value | number | 3,412,000 | The central forecast, in the unit of the country's headline series. |
low | number | 3,198,000 | Bottom of the range at the level asked (80 by default, 95 with level=95). |
high | number | 3,641,000 | Top of the same range. |
meta.method.backtest_mape_pct | number | 7.6 | Mean absolute error, in percent, when the last 12 published months were forecast blind. |
meta.method.fitted_from | string | 2022-12 | The first month the model learned from: the reopening, not the pandemic. |
meta.method.horizon_months | number | 24 | How far ahead it runs. |
meta.method.made | string | 2026-10-20 | The day it was fitted. |
meta.method.method | string | Holt-Winters on logs, damped slope 0.9, season started from 2017 to 2019 | Holt-Winters on logs, damped slope 0.9. |
3 in all.
| Parameter | Values | Default | Use |
|---|---|---|---|
country | ISO code | required | One country per call. |
horizon | 1 to 24 | 24 | Months ahead to return. |
level | 80, 95 | 80 | The range in low and high. |
JSON by default, CSV with format=csv.
# Forecast for Japan curl "https://api.tourismapi.com/v1/stats/forecast?country=JP" \ -H "X-API-Key: $TOURISM_API_KEY"
From the nightly build. The first three rows are open; 70 more come with Intelligence.
| Country | Last 12 months | Next 12 months | Change | 80% range | 95% range | Back-test error | Fitted from | Actions |
|---|---|---|---|---|---|---|---|---|
| Italy | 88.5m | 577.3m | +552.5% | 465.4m to 719.5m | 416.0m to 809.8m | 4.6% | 2022-01 | |
| Spain | 81.2m | 527.2m | +548.8% | 465.6m to 598.1m | 436.3m to 640.0m | 3% | 2022-01 | |
| Germany | 37.6m | 445.3m | +1,082.9% | 388.8m to 511.1m | 362.2m to 550.3m | 1.9% | 2022-01 |
Four buyers, four calls.
Set next season's rates against the forecast and its range.
horizon=12Size staff and rooms to the top of the 95% range.
level=95Weigh new routes by where arrivals are heading.
horizon=24Quote the forecast with its error, never without.
backtest_mape_pctThe statistics that answer the next question.
Every plan reads the same figures; this statistic comes with Intelligence and up.
| Free | Developer $49 | Analyst $99 | Intel $199 | |
|---|---|---|---|---|
| Forecast | not included | not included | not included | included |
| The figures it is worked from | a year back | ten years, latest month | full history | full history |
| Trend, recovery, benchmark, stay length | not included | not included | included | included |
| Forecast, nowcast, anomalies, origin shift, concentration | not included | not included | not included | included |
| Calls a month | 1,000 | 50,000 | 250,000 | 1,000,000 |
| Calls a minute | 10 | 60 | 120 | 300 |
Includes all nine statistics, and everything in the plans below it. Ten months for the price of twelve when paid yearly. Cancel any time.