open-climate.aiDocs

Export to a spreadsheet

Numbers checked against the live data on

To export, set your filters and columns in factor search, click Export, choose CSV or XLSX, and click Export again. The file holds every factor your filters match, with the columns you see on screen. Browsing and factor pages are free; the export needs a free account.

Before you start#

  • An account. Without one, the dialog still opens and tells you what the file would hold, under Free with an account, with Create free account and Already have one?. Signing up brings you back to the same view, with your search and filters. It is the same free account that lifts the search allowance.
  • Your filters. The export follows the filters in the Filters rail, including the Year filter. Find a factor shows how to set them.
  • Your columns. Only the visible columns are exported. The Columns menu says it at its foot: "Visible columns are the exported columns."

Export the view#

  1. In factor search, set the filters you need.
  2. Click Columns and show the columns you want in the file, for example Approach, Grid mix or Data quality. Hide the ones you do not need.
  3. Click Export above the list, or press E. The dialog Export this view opens.
  4. Check the summary under the title: the number of factors and the number of columns, "(as shown)".
  5. Under Format, choose CSV or XLSX.
  6. Click Export. The dialog reads "Preparing your file…", then names the file it downloaded.

On a library page, the Browse factors section has the same Export button, already limited to that library.

What the file contains#

  • The rows your filters match, with the values factor search shows.
  • The visible columns, in the order on screen. The Year column becomes two columns in the file, the start and the end of the year window, because a cell cannot hold a range.
  • The licence of each row. Every file also carries the licence code and the credit line of each row, even if those columns are hidden on screen. The dialog says that the source terms apply per row. Keep these columns when you pass the file on: see Cite a factor.

An empty cell is a value the publisher does not give. It is not zero. Do not fill it with 0 when you add up the file. See value.

What the columns mean#

The header row carries each column's machine name, not the label you saw on screen: Activity is activity_full, GWP is gwp_method, Derived is is_derived. Those are the same names the reference tables use, so a file and a query can be read against each other. A yes-or-no column reads true or false.

in the file on screen what the cell holds
activity_full Activity What the factor measures, in the publisher's own words. See activity.
library Library The published dataset the number comes from. See library.
category, sub_category Category, Sub-category Where the factor sits on our category axis, which is the same tree for every library. See category.
region_code Region The geography the factor describes. See region.
system_boundary, boundary_family Boundary, Boundary family Which life-cycle stages the value covers, and the taxonomy that boundary code was defined in. For how much it covers, keep the Boundary column and read boundary_preset and boundary_role beside it. See boundary.
applies_from_year, applies_to_year Year The first and last accounting year the value applies to. See year window.
value_co2e Value Kilograms of CO2e per one unit, on the warming potentials named in gwp_method on the same row. See value.
unit_code, unit_family Unit, Unit family The denominator the value is per, and the kind of quantity it is. See unit basis.
co2e_basis CO2e basis Which carbon the CO2e counts: fossil only, biogenic excluded, a net total, or an input-output total. See CO2e basis.
gwp_method GWP The IPCC report whose warming potentials converted each gas. See GWP methodology.
calculation_approach Approach For electricity, location_based or market_based. See calculation approach.
radiative_forcing Radiative forcing For a flight, whether the extra warming at altitude is included. Empty for anything that is not a flight. See radiative forcing.
grid_mix Grid mix Which electricity a kWh is assumed to be. See grid mix.
price_basis, currency_year Price basis, Currency year For a spend factor, which price the money is and the price level it is in. See price basis and currency year.
data_quality_normalized Data quality The publisher's own rating put on one scale. See data quality.
is_derived Derived true when the total was assembled from parts the publisher printed separately. See derived total.
supplier_name, supplier_code, has_supplier Supplier, Supplier code, Named supplier The counterparty the factor is specific to, when it names one. See supplier.
data_type Data type Whose number it is: specific, average or generic. See data type.
is_epd EPD true for a factor read off an Environmental Product Declaration. See EPD.

Watch out. Nothing in this list identifies the factor itself. The file factor search writes has no slug column, no source_key, no release version and no export date, so on its own it cannot be matched back to factor search later. Keeping an export checkable says what to record, and how to get the identifiers into the file.

The columns you did not ask for#

Three groups of columns appear in the file without a tick in the Columns menu.

  • The licence pair, license_code and attribution_text, on every row, whatever you chose. They are the terms and the credit line that travel with the row: see Cite a factor.
  • The window start, applies_from_year, whenever you keep the Year column. That column is the window's last year, applies_to_year; one cell cannot hold a range, so the file writes both ends, start first.
  • The coverage columns, whenever you keep the Boundary column. phase_upstream, phase_production, phase_downstream and phase_eol each say how much of that life-cycle phase the number covers, boundary_role says whether the number is a whole answer, one part to add to others, or a credit to keep apart, and boundary_preset is the reader-facing group the Boundary filter offers. They restate one boundary code as the four questions everyone shares, so you can compare coverage across libraries without learning the code list. See System boundaries.

A phase reads full, partial or none where the coverage is known, unstated where the publisher is silent, and n_a where the product has no such phase, such as the end of life of a litre of fuel. Read unstated and n_a as unknown and not applicable, never as zero.

If you typed a search, the dialog warns you: the text search is not exported, because a file has no ranking. The file holds every row your filters match, which is usually many more rows than the ranked matches on screen.

To export only what you need, turn your search into filters first: click the suggested chips under the search box, or pick Category, Region and Unit in the rail, then Clear search.

Rows left out of the file#

A library whose licence does not let us republish its values has those values withheld: a dagger (†) in factor search, and no row in the file. Before you export, the dialog counts them for you, in the form "3 licence-restricted rows are not included in the file."

What is left out follows the licence tier, not a list of names: open and copyleft libraries are served in full, restricted ones are searchable with their values withheld. The Licence filter in the rail tells you which tiers your view holds, and the Open data page marks a restricted library "Searchable on the portal · values withheld".

Large exports#

When the view is too large to build at once, the dialog queues it instead. It shows the progress in rows, then "Your file is ready." and a Download file link. The link lasts seven days; after that, run the export again.

The same export over the API#

Once you are signed in, the dialog also shows The same export over the API: the web address that produces the same file, with Copy URL. Give it to a developer who wants to fetch the file on a schedule; you do not need it to export.

Keeping an export checkable#

A file you took keeps the numbers it was cut with, whatever factor search serves afterwards: that is what a refresh changes, and it is the point of exporting. It is also the problem. Six months later, "has any of this changed?" is a question you can only answer by looking each row up again, and the columns factor search writes give you nothing to look them up by. Matching rows back on their activity text is slow, and ambiguous whenever two factors describe the same activity.

Two things keep an export checkable, and both are decided before you export.

  • Record what the file cannot hold. The date you exported, the filters you set, and the edition of each library the view covers, which the Open data page prints on every card. Those three say which data the file was cut from, and nothing on the site writes them down for you. A Dataset date copied from a factor page's citation is not a stand-in: it stamps the site build, not the data. See what the Dataset date is.
  • Or put the identifiers in the file, over the API. The Columns menu cannot offer slug and source_key, because the columns you can see are the columns you can export. An export asked for over the API can name them: copy the address under The same export over the API and ask a developer to add columns=slug and columns=source_key to it. Everything else about the file stays as it is, licence columns included.

What the API will name is the served row: the columns in What the columns mean, the identifiers, and the per-gas and licence fields beside them. It is not every column of the SQL model, and a name outside the set is refused rather than ignored, with "Unknown export columns". Two fields that Pin a factor value to a release asks you to keep, price_index_ratio and value_key, are on that far side: both are computed when the model's served view is built, so a query is the only place to read them.

Later, slug and source_key are what you compare. The slug is permanent, and the source_key beside it changes when the served value does, whether from a new edition, a newly elected default or a correction. Pin a factor value to a release writes that comparison out, including the two served numbers a source_key does not pin, and A source_key never changes is the promise it rests on.

Other ways to take data away#

  • One factor. Factor pages have an Export button that is not active yet. Copy the value and its citation from the page instead: see Cite a factor.
  • A whole library. The library page offers the publisher's edition as one file: see Download the open data.

Next#