Transforms

Add ?transform= to /v1/series/{id} and the API computes the transform for you. Your numbers then match everyone else's, and you don't have to fetch extra history to do it yourself.

curl -H "X-API-Key: pk_live_xxx" \
  "https://api.pakdatahub.com/v1/series/inflation.cpi.national?transform=yoy&limit=2"

{ "success": true, "series": "inflation.cpi.national",
  "meta": { "name": "CPI National", "unit": "percent", "frequency": "monthly",
            "source": "PBS", "last_updated": "2026-06-30", "transform": "yoy" },
  "data": [ { "date": "2026-06-30", "value": 11.090083270249806, "dims": {} },
            { "date": "2026-05-31", "value": 11.64643399089529, "dims": {} } ],
  "pagination": { "next_cursor": null, "count": 2 } }

#Rules

• Partners are fetched for you. yoy, mom, pct_change and 3ma look back before your from date and outside your limit, so every returned point has its partner. A point is null only where no earlier observation exists at all (e.g. the first year of a series).
• Per dimension. Transforms run separately for each dims group (KIBOR bid and offer, each SPI city).
• index uses your window. It rebases to 100 at the first point you get back, so pick from deliberately:

# Rupee depreciation since 2018, as an index
curl -H "X-API-Key: pk_live_xxx" \
  "https://api.pakdatahub.com/v1/series/fx.rate.avg.usd?from=2018-01-01&transform=index&sort=asc"

• Mismatches fail loudly. mom on a quarterly series, or yoy on an annual one, returns 422 invalid_params with a message suggesting pct_change.
• Transforms combine with vintage, dims, format=csv and sort.

#Recipes