Skip to content
This repository has been archived by the owner on Apr 30, 2022. It is now read-only.

Latest commit

 

History

History
346 lines (237 loc) · 11.3 KB

FOR_DEVELOPERS.md

File metadata and controls

346 lines (237 loc) · 11.3 KB
Raw

Detailed Method Guide - Quandl/Python

In addition to the Quick methods for retrieving data, some additional commands may be used for more querying specificity. These include:

  • Retrieving metadata without data
  • Customizing how data is returned more granularly
  • Allowing easier iteration of data

In each of the following sections it is assumed your Quandl API key has been set via:

import quandl
quandl.ApiConfig.api_key = 'tEsTkEy123456789'

Retrieving Data

In the following sections, params={} represents optional query parameters that can be passed into each call. For more detail on available query parameters please see the API Documentation.

Dataset

A dataset's data can be queried through the dataset object. For example:

data = quandl.Dataset('WIKI/AAPL').data()

A number of optional query parameters can be passed to data():

dataset_data = quandl.Dataset('WIKI/AAPL').data(params={ 'start_date':'2001-01-01', 'end_date':'2010-01-01', 'collapse':'annual', 'transformation':'rdiff', 'rows':4 })

You can access the data much like you would other lists. In addition all the data columns, fields are mapped to their column_names for convenience:

dataset_data[0].date

Datatable

A datatable's data can be retrieved in much the same was as a dataset. For example:

data = quandl.Datatable('ZACKS/FC').data()

Note that unlike a dataset, datatables information may be paginated. If the data object returned back contains a cursor_id you will need to make another call appending that cursor_id to the datatables parameters.

data2 = quandl.Datatable('ZACKS/FC').data(params={'qopts': {'cursor_id': data.meta['next_cursor_id']}})

Note that all parameter options are given under the kwarg params. These parameters will be passed through the API call. You can see a complete list of parameters here.

We also recommend using filters. This can help to reduce the number of pages returned in the result set. For example:

data = quandl.Datatable('ZACKS/FC').data(params={'ticker': ['AAPL','MSFT'], 'per_end_date': {'gte': '2015-01-01'}, 'qopts': {'columns': ['ticker', 'comp_name']}}, qopts={'columns':['ticker', 'per_end_date']})

This however may still result in multiple pages so be sure to check for a resulting cursor_id. An example code that retrieves all pages of data with a filter may look something like:

data_list = []
cursor_id = None
while True:
    data = quandl.Datatable('ZACKS/FC').data(params={'ticker': ['AAPL','MSFT'], 'per_end_date': {'gte': '2015-01-01'}, 'qopts': {'columns': ['ticker', 'comp_name'], 'cursor_id': cursor_id}})
    cursor_id = data.meta['next_cursor_id']
    data_list.append(data)
    if cursor_id is None:
        break

Download table data as a zip file. You can download all the table data in a data table in a single call. The following will download the entire F1 table data as a zip file to your current working directory:

import quandl
data = quandl.export_table('MER/F1')

You can also specify where to download the zip file:

import quandl
data = quandl.export_table('MER/F1', filename='/my/path/db.zip')

Note that if you are downloading the whole table, it will take longer to generate the zip file.

You can also specify what data you want to download with filters and parameters.(cursor_id and paginate are not supported for exporting table zip file):

import quandl
quandl.export_table('ZACKS/FC',  ticker=['AAPL', 'MSFT'], per_end_date={'gte': '2015-01-01'}, qopts={'columns':['ticker', 'per_end_date']})

After the download is finished, the filename of the downloaded zip file will be printed.

Sometimes it takes a while to generate the zip file, you'll get a message while the file is being generated. Once the file is generated, it will start the download of the zip file.

Download Entire Database (Bulk Download)

To get the url for downloading all dataset data in a database:

quandl.Database('ZEA').bulk_download_url()
=> "https://www.quandl.com/api/v3/databases/ZEA/data?api_key=tEsTkEy123456789"

To bulk download all dataset data in a database:

quandl.Database('ZEA').bulk_download_to_file('/path/to/destination/folder_or_file_path')

For bulk download of premium databases, please ensure that a valid api_key is set, as authentication is required.

For both bulk_download_url and bulk_download_to_file, an optional download_type query parameter can be passed in:

quandl.Database('ZEA').bulk_download_to_file('.', params={'download_type': 'partial'})

If download_type is not specified, a complete bulk download will be performed. Please see the API Documentation for more detail.

MergedDataset

You can get a merged representation of multiple datasets.

merged_dataset = quandl.MergedDataset([('WIKI/AAPL', {'column_index': [11]}),
                                           ('WIKI/MSFT', {'column_index': [9,11]}), 'WIKI/TWTR'])

In the above example the following data will be merged together:

  • column 11 of dataset 'WIKI/AAPL'
  • columns 9 and 11 of dataset 'WIKI/MSFT'
  • all columns of 'WIKI/TWTR' is requested

To get the data for the MergedDataset:

data = merged_dataset.data()

The same optional query parameters shown in Dataset data can be passed to data()

data represents a full outer join of data from the requested datasets. data can also be converted to csv, pandas, and NumPy in exactly the same way as Dataset data

Data formats

To convert the data into csv format:

data.to_csv()
=> "Id,Name,Database Code,Description,Datasets Count,Downloads,Premium,Image,Bundle Ids,Plan ...

To convert the data into a pandas dataframe:

data.to_pandas()

To convert the data into a NumPy record:

data.to_numpy()

Since the data is a List, the raw data can be retrieved via:

data.to_list()

All options beyond specifying the dataset WIKI/AAPL are optional.

See the pandas and NumPy documentation for a wealth of options on data manipulation.

Point in Time

Point in time data can be retrieved in much the same was as a datatable. For example:

data = quandl.PointInTime('DATATABLE/CODE', pit={'interval': 'asofdate', 'date': '2020-01-01'}).data().to_list()
data = quandl.PointInTime('DATATABLE/CODE', pit={'interval': 'asofdate', 'date': '2020-01-01'}).data().to_pandas()
# or
data = quandl.PointInTime('DATATABLE/CODE', pit={'interval': 'from', 'start_date': '2020-01-01', 'end_date': '2020-01-15'}).data()

For more options please check this table

Retrieving metadata

Dataset

To retrieve metadata about a dataset simply instantiate its object using its Quandl code:

quandl.Dataset('WIKI/AAPL')

Once instantiated you can then make data or metadata calls on the object. A metadata call looks like:

quandl.Dataset('WIKI/AAPL').data_fields()
=> ['premium', 'name', 'frequency', 'description', 'column_names', 'database_code', 'type', 'refreshed_at', 'newest_available_date', 'dataset_code', 'oldest_available_date', 'database_id', 'id']

Note that a call to any attribute such as name will trigger an API metadata call if the metadata has not been loaded yet.

Database

To retrieve metadata about a database simply instantiate its object with a database code and then call a metadata method on it:

db = quandl.Database('WIKI')
db.name

You can also get data through other objects such as a dataset:

dataset = quandl.Dataset('WIKI/AAPL')
dataset.database()

Retrieve a list of databases by using:

quandl.Database.all()

By default, each list query will return page 1 of the first 100 results (please see the official API Documentation for more detail).

Retrieve the dataset through the database by using the helper method.

quandl.Database('WIKI').datasets()

Datatable

Much like databases and datasets you can retrieve datatable metadata via its Quandl code:

dt = quandl.Datatable('ZACKS/FC')
dt.data_fields()

Working with results

Instance

All data once retrieved is abstracted into custom classes. You can get a list of the fields in each class by using the data_fields method.

database = quandl.Database('WIKI')
database.data_fields()
=> ['name', 'downloads', 'id', 'premium', 'description', 'datasets_count', 'database_code', 'image']

You can then uses these methods in your code. Additionally you can access the data by using the hash equalivalent lookup:

database = quandl.Database('WIKI')
database.database_code
=> 'WIKI'
database['database_code']
=> 'WIKI'

In some cases the names of the fields returned by the API may not be compatible with the Python language syntax. These will be converted into compatible field names.

data = quandl.Dataset('WIKI/AAPL').data(params={ 'limit': 1 })[0]

data.column_names
=> ["Date", "Open", "High", "Low", "Close", "Volume", "Ex-Dividend", "Split Ratio", "Adj. Open", "Adj. High", "Adj. Low", "Adj. Close", "Adj. Volume"]

data.data_fields()
=> ["date", "open", "high", "low", "close", "volume", "ex_dividend", "split_ratio", "adj_open", "adj_high", "adj_low", "adj_close", "adj_volume"]

List

All list queries will return an object inherited from ModelList.

To get the values of a list:

databases = quandl.Database.all()
databases.values

To get the metadata of a list:

databases = quandl.Database.all()
databases.meta

To get the raw data of a list:

databases = quandl.Database.all()
databases.to_list()

Database.all() and Dataset.all() will return a paginated list of results in the form of a PaginatedList object. You can check whether the PaginatedList has more data by using the has_more_results() method. By default, each list query will return page 1 of the first 100 results (please see the official API Documentation for more detail). Depending on the list query results, you can pass additional parameters to filter the data:

databases = quandl.Database.all()
=> ... results ...
databases.has_more_results()
=> true
quandl.Database.all(params={ 'page': 2 })
=> ... more results ...

Lists also function as arrays and can be iterated through. Note however that using these features will only work on the current page of data you have locally. You will need to continue to fetch results and iterate again to loop through the full result set.

databases = quandl.Database.all()
for database in databases:
    print(database.database_code)
=> ... print database codes ...
databases.has_more_results()
=> true
more_databases = quandl.Database.all(params={ 'page': 2 })
for database in more_databases:
    print(database.database_code)
=> ... print more database codes ...

Lists also return metadata associated with the request. This can include things like the current page, total results, etc. Each of these fields can be accessed through a hash or convenience method.

quandl.Database.all().current_page
=> 1
quandl.Database.all()['current_page']
=> 1