How every score is worked out

Are visits spread across the country, or piled into a few states? 0 = every state gets the same. 1 = one state gets everything.

1. 1Sort the 16 states from fewest visitors to most.
2. 2Give them weights from −15 (smallest) up to +15 (biggest), in steps of 2.
3. 3Multiply each state's visitors by its weight and add everything up.
4. 4Divide by (16 × total visitors).

Which states have a lot to offer but few visitors? Opportunity = what it can offer − how visited it already is.

1. 1Score every ingredient from 0 to 100: (state − lowest state) ÷ (highest − lowest) × 100.
2. 2Counts (rooms, attractions, visitor measures) are logged first, so Kuala Lumpur's size does not flatten everyone else.
3. 3Offer = average of 7 scores. Visited = average of 3. Equal weights unless you move the sliders.
4. 4Opportunity = Offer − Visited. Above 0 is untapped, below 0 is crowded.

What holds a state back: getting there (access), being known (awareness), or places to stay (amenities)? Four ingredients each.

1. 1For each ingredient, find the typical state: the median of the 16.
2. 2Typical spread = the median distance from that median × 1.4826.
3. 3Distance = (state − median) ÷ typical spread, limited to −3 … +3.
4. 4Pillar score = 50 + (average of its 4 distances) × 50 ÷ 3. 50 = the typical state.
5. 5Bottleneck = the lowest pillar, flagged only if it is below 50.

How many more visitors can the hotels that already exist take, without building anything?

1. 1Spare room-nights = rooms × 365 × (75% − occupancy). The ceiling is yours to change in the Simulator.
2. 2Room-nights one visitor needs = overnight share × paid accommodation share × nights ÷ 2 per room.
3. 3Extra visitors = step 1 ÷ step 2.

What changes if some visitors go to another state instead? A what-if calculator, not a forecast.

1. 1Visitors moved = the crowded state's visitors × the chosen %.
2. 2Split them across the destinations. None takes more than its room to grow; the overflow goes to the others.
3. 3Gained = moved × destination's spend per visitor. Lost = moved × origin's spend per visitor. Net = gained − lost.
4. 4Gini, top-3 share and hotel occupancy are recalculated.

How do real travellers describe each state? The only part that uses AI, and only to label text.

1. 1A ready-made model (Gemini 2.5 Flash Lite) tags each post from fixed lists: real trip or not, place, topic (1 of 11), negative / mixed / positive.
2. 2The "real trip?" question is asked twice; a post is kept only on two yeses: 7,561 posts.
3. 3Negative = 0, mixed = 50, positive = 100. Average per state and topic.
4. 4Few mentions? The average is pulled towards the national average: (mentions × state average + k × national average) ÷ (mentions + k).
5. 5Experience Score = average of the adjusted topic scores.