APIxMinistry of Statistics · Problem SIH26056

Every fare in India,
reduced to one honest number

The rule on record prices airfare for the CPI once a month, at a single fifteen-day advance booking. APIx prices its whole basket every day, across five advance-purchase windows, and reports what it could not match.

—

Routes weighted

1×

sweep per day

—

Fares collected

—

Index

See the price surface
What this page shows. One number for what it costs to fly in India right now, against the average of the reference window it is based on. It is built from — fares collected automatically across — routes, once a day, at five advance-purchase windows, with no one typing anything in.
National airfare price index
—
— vs previous day— vs reference window

Read it like this: flying costs — than it did over the five days the index is referenced to.

— · base = 100 · departure-date basis · Jevons × Young

Proof attached to this number
Fares used—
Coverage—
Estimated—
Revised sincenever
Transitivity drift—
Index awaiting its first complete reference window

The dark line is the headline index. Switching to booking windows splits it into the five advance-purchase leads it is built from (T+1 through T+45), which is where most of the day-to-day movement actually lives.

APIx
Routes weighted—DGCA passenger-traffic basket
Checks per day1the rule on record: one a month
Fares read daily—no human in the loop
How stale—since the last completed sweep
Airfare in CPI—its weight in the 2024 basket
Where the basket reaches

Click a hub or a corridor chip to inspect it. The moving beacon is a visualization of the selected corridor, not live flight tracking. The index, cells priced and basket weight shown for it are real, pulled from the same data as the table above.

SPACE VIEW · STATES & BORDERS DEMARCATED
20.59° N, 78.96° E · ORBIT 480 KM · visualization, not live tracking
12 BASKET CORRIDORS
Satellite view of India with the airport hubs in the airfare basket
JAMMU & KASHMIRLADAKHPUNJABHARYANARAJASTHANUTTAR PRADESHGUJARATMADHYA PRADESHBIHARWEST BENGALMAHARASHTRACHHATTISGARHODISHATELANGANAANDHRA PRADESHKARNATAKATAMIL NADUKERALAASSAMDELBOMPNQBLRHYDCCUMAAAMD
Selected corridor (1 of 12)
BLR ⇄ DELNo data yet
Jevons index
—
Cells priced
—
Basket weight
—
Click a corridor chip or a hub to inspect
Which routes moved the index today

Busy routes move the national number more than quiet ones, because each route is weighted by expenditure (passengers times mean fare), not by how many flights it has.

Pushed the index upPulled it down
These do not add up, and that is the point. The twelve contributions sum to — against a headline move of —. The gap is imputed cells and day-to-day sample churn, and it is shown rather than spread silently across the routes.
The number leaves as plain, versioned JSON

No integration project and no bespoke format. Every figure on this page is one public GET away, with its coverage and reference window attached to the response.

GET /api/v1/index

{
  "series_id": "APIX.ALL",
  "reference": { "factor": …, "is_provisional": … },
  "coverage":  {
    "routes_with_data": …, "routes_in_basket": …,
    "n_points": …
  },
  "points": [ …, {
    "period_start": "…",
    "level": …, "n_cells": …,
    "n_cells_imputed": …, "weight_imputed": …
  } ]
}
Nothing is published without its shortfall. Every response carries how many routes reported, how much was imputed, and whether the reference window is still provisional, so a number can be checked before it is quoted.