Using the Quandl Website
This document is a comprehensive guide to using the Quandl website. It teaches you how to find the data you need; how to get the data in the format you want; and how to make the best use of Quandl's many data-handling features.
Datasets and Sources
The basic unit of data in Quandl is the “dataset”. Here are some examples of datasets:
- GDP of the United States
- Inflation in China
- AAPL stock price
- Kansas City wheat futures
- Deforestation in Brazil
Quandl currently handles only time-series datasets, i.e., datasets where the first column is a date. We don’t handle intraday data yet. Nor do we handle non-numerical data.
Each dataset on Quandl is associated with a “source”. Sources can be data publishers, curators, contributors, vendors or individual users. Here are some examples of popular sources on Quandl:
Each dataset on Quandl has a unique "Quandl code", comprising a source code and a table code. For instance, the dataset named GDP of the United States has the Quandl code FRED/GDP, where FRED is the source code and GDP is the table code. All datasets from the same source will have the same source code.
There are several ways to find datasets on Quandl. The first is to use “Search”.
Quandl search works much like Google or any other search engine. Type in a word or phrase that describes what you’re looking for, and Quandl will try to return the most relevant datasets for your query. Quandl looks at dataset names, descriptions, column headers, Quandl codes, source names, hidden keywords and popularity when determining relevance.
Individual search results on Quandl include rich metadata about each dataset returned:
You can use this metadata to decide which dataset(s) you're interested in.
More About Search
By default, Quandl searches all data sources and all data frequencies. But you can narrow down your search results using the filters on the left. If the source you are interested in does not appear in the filter list, simply type its name or source code into the “Add Source” text box, and Quandl will add it to the list.
Quandl limits its search results to display at most two results from any one source. To view more results from a particular source, click on “Show all results from [source]”. (Note that this is equivalent to selecting that particular source in the filter list on the left).
In addition to dataset results, Quandl shows a few "collection"" results. Collections are groups of datasets on specific topics, hand-selected by Quandl curators for quality, accuracy and relevance. In many cases, collections will lead to better (i.e. more useful) datasets than basic search.
You can see all the collections pertaining to your search query by clicking on “Show more collections”.
Sometimes basic search is not sufficient to find the dataset you need. In that case, you can use Quandl’s “Advanced Search”.
Clicking on “Advanced Search” loads a new window where you can customize your search query. Note that you can access the same functionality within Quandl’s main search interface, using the "AND", "OR" and "NOT" operators.
For example, searching for:
crude OR refined AND oil NOT canola
will return results that definitely contain the word "oil", may contain either, both or none of the words "crude" and "refined", and definitely do not contain the word "canola".
In place of "NOT", you can also use the minus operator "-".
You can search for datasets with specific table codes by prefacing the string "code:" to your query. For example, searching for:
will return datasets whose Quandl codes exactly match the string UNRATE.
You can use the wild-card operators "?" and "*" anywhere in your query string. This is especially useful when searching for datasets by partial code match. For example, searching for:
will return all datasets whose Quandl codes contain the string YBHA.
Each and every dataset indexed by Quandl has its own page. "Dataset Pages" have unique URLs in this format:
So if you know a dataset’s Quandl code, you can quickly jump to its dataset page on quandl.com.
Quandl dataset pages share a common format and appearance:
The green tutorial button shows a very brief animated walk-through of the main features on the data page; do click it!
The next few sub-sections describe how to navigate and use various features on the dataset page.
You can perform the following actions using the data toolbar:
Change Date Range
By default, Quandl shows all the data available for any dataset. You can use this control to trim the date range: show everything, or just a subset of the data.
If you have (say) a daily dataset, you can use this control to convert it to a monthly, quarterly or annual dataset. Similarly, you can collapse monthly data to quarterly or annual. But you cannot go in the opposite direction: you cannot convert annual data to monthly or daily.
Note that Quandl always takes the “last available” high-frequency observation when converting from high frequency to low. Thus this function does not work well for datasets that measure percentage changes, period averages or period extremes (highs and lows).
More About Frequency
"Default Frequency" means date stamps are precisely what the original publisher used. You can change the frequency to anything you want. When you select a frequency, Quandl does two things:
It generates all the dates that should appear between start date and end date given the frequency you have chosen. So, for example, if the data is daily and goes from January 15, 2012 to August 19 2012 and you select "monthly" frequency, then Quandl generates the dates: Jan 31, Feb 28, March 31, ... July 31, Aug 19.
Quandl then selects the last observation for each period as the data value for that month. So, if there is an observation on January 31, it will be used as the January data point. If the last observation in January occurred on January 28th, then that will be the data point used for January. The last observation for August occurs on the 19th, so that will be the value used for August.
Question: What happens if the dataset is originally monthly and I change the frequency to monthly?
Answer: The process above is still followed. This means that you will now see when data is missing. If there is actually no data point for June 2007 you might not see a gap in the original data. On one line you will see May 2007 and on the next you will see July 2007. But when you choose monthly as the frequency, missing data points will become apparent. You will see a date stamp of June 2007 and the data will be noticeably missing.
Use this control to carry out elementary transformations on your data: row-on-row change, percentage change, cumulative sum, and normalize (set starting value at 100).
Click on the red "Download" button to get the raw data directly onto your computer, in whatever format you want. Formats currently offered include CSV, JSON and XML. (If there's some other format you'd like us to implement, please don't hesitate to email us).
Note that you don't necessarily need to download raw data prior to anlayzing it. Quandl offers integrated "packages" for many third-party analysis tools, including Python, R, Excel, Matlab, Stata and more. Using these packages, you can get the data you need directly into the tool of your choice. Visit our packages help page to learn more.
In addition to packages, the data download window also shows the full API call corresponding to your dataset. You can thus access this data directly from your own app, should you choose. Visit our API help page to learn more.
Note that the downloaded data will reflect whatever X-axis date range, frequency and transformation you have previously selected using the other controls on the data toolbar.
Click on the blue "Embed" button to launch a wizard that will help you embed a Quandl graph on your own website, blog or comment thread.
The graph embed wizard allows you to select a graph title, a source attribution, a range for the Y-axis, and a graph size in pixels. (All of these fields are optional; if you leave them blank, Quandl will fill them in for you). Simply copy and paste the "embed code" into your website or blog, and a graph will appear at that location.
Note that the embedded graph will reflect whatever X-axis date range, frequency and transformation you have selected using the other controls on the data toolbar.
Advanced tip: Omit the "trim_end" parameter in the embed code to create an embedded graph that updates "live" from the source -- each time the data on Quandl changes, the graph will update automatically.
Use this button to report problems with a dataset: missing data, bad data, attribution errors, incorrect or incomplete descriptions, or any other issues you might spot. Quandl relies on user contributions for its coverage and accuracy; thank you for taking the trouble to report data problems!
The green "Tutorial" button shows a very brief visual guide to using Quandl's dataset pages.
You can customize the graph in the following ways:
Select Series To Plot
For datasets with more than one data column (such as NSE/OIL but not FRED/GDP) you can select which column to plot by simply clicking on the column name, to the right of the graph. You can plot as many columns as you want, at the same time.
Select Date Range
You can change the date range for display (i.e. the range of the X-axis on the graph) using the slider control below the graph. Note that changing the display range in this manner does not alter the date range for download; for that you have to use the date control in the toolbar.
View Full History
Double click anywhere on the slider below the graph to view the full history for your dataset (i.e. to expand the date range to its maximum).
Hover your mouse over any point on the graph to see the exact date and value for that point.
The dataset page also has the following sections:
This is what it's all about! Here's the data you need, updated fresh from the source, and presented to you in a canonical format irrespective of how and where it was originally published.
Advanced tip: Click on any column header (including the date column) to sort the entire table on that column. Click again to toggle the sort order.
This has useful metadata about the dataset you're looking at, including: how frequently the source updates, how recently Quandl retreived data from the source, the name of the source, a link to help validate the data, a permalink to the data on Quandl, the dataset's Quandl code, and a brief description.
To add a dataset to your favourites, simply click on the gold star next to the dataset name. Note that you need to be signed in in order to use this feature.
To view all your favourite datasets, click on "Favourites" in the top navigation bar (next to your username). This reveals a dropdown list of all your favourites. You can quickly jump to a particular favourite by typing keywords into the special "favourites search box" on the top right of this dropdown. You can also view all your favourites by clicking on "Manage Favourites" at the end of the dropdown.
To remove a dataset from your faourites, simply uncheck the gold star on that dataset's page. Alternatively, go to "Manage Favourites" under the "Favourites dropdown menu" and uncheck the star there.
You need to be signed in to your Quandl account in order to use favourites. Getting a Quandl account is completely free and takes just a few seconds! Visit our accounts help page to learn more.
You can merge any 2 or more of our 8,000,000+ datasets to create a brand-new, persistent dataset that updates automatically when its consituents update. We call these merged datasets "supersets".
Supersets function just like any other dataset: you can transform, collapse, trim, download, embed, or API-access a superset just like you would any other Quandl dataset.
To create a superset, first locate the specific data column that you want to add to your superset. Click the black table icon next to the column name. You will get a popup with the words "Add to Superset". Hovering over this popup reveals a list of existing supersets that you can add this column to; alternatively, at the bottom of the list is an option to add the column to a "New Superset". Select whichever option you prefer.
If you chose "New Superset", Quandl will create a superset with the name "Untitled Superset [Date and Time of Creation]" for you. Note that if you're creating a new multi-column superset from scratch, you should only have to select "New Superset" once; after that, you will be adding extra columns to the Untitled Superset that Quandl just created for you.
Add as many columns to your superset as you'd like. Each time you add a column to a superset, Quandl shows a green status bar with a link to that superset. Click on that link to view the superset's dataset page.
You need to be signed in to your Quandl account in order to create supersets. Getting a Quandl account is completely free and takes just a few seconds! Visit our accounts help page to learn more.
You can change your superset's name at any time; simply click on the name to edit it. Similarly, you can edit the superset's description field by clicking on it. And you can rename (or remove) any column in your superset by clicking on the black table icon next to the column name.
Supersets have Quandl codes in which the source code is your username (in ALL CAPS) and the table code is automatically generated by Quandl. You cannot edit a superset's Quandl code.
Supersets can be public or private. Public datasets are visible to anyone who has the URL (or equivalently, anyone who knows the Quandl code); however, they do not appear in website search results, nor are they indexable by Google. Private datasets are visible only to you, and only when you're signed in. Use the privacy toggle to choose which option you want; the default is "public".
Here's a one-minute video showing how to create, share and re-use a superset on Quandl:
Collections are groups of related datasets on specific subjects curated by Quandl users. Here are some examples of popular collections on Quandl:
Collections offer an easy way to find high-quality data on the subject of your interest. You can browse through all of our 75,000+ data collections from our home page.
Quandl is best accessed by the Chrome, Firefox, and Safari browsers; for Internet Explorer it works with version 9 and higher. If you're not using a recommended browser and are experiencing issues throughout the site it may be due to incompatibilities.
If you are still encountering issues, please email us at firstname.lastname@example.org.