gaapi4py

Google Analytics Reporting API v4 for Python 3

Prerequisites

To use this library, you need to have a project in Google Cloud Platform and a service account key that has access to Google Analytics account you want to get data from.

Quick Start

from gaapi4py import GAClient
# if GOOGLE_APPLICATION_CREDENTIALS is set:
c = GAClient() 
# or you may specify keyfile path:
c = GAClient(json_keyfile="path/to/keyfile.json")


request_body = {
    'view_id': '123456789',
    'start_date': '2019-01-01',
    'end_date': '2019-01-31',
    'dimensions': {
        'ga:sourceMedium',
        'ga:date'
    },
    'metrics': {
        'ga:sessions'
    },
    'filter': 'ga:sourceMedium==google / organic' # optional filter clause
}

response = c.get_all_data(request_body)

response['info'] # sampling and "golden" metadata

response['data'] # Pandas dataframe that contains data from GA

If you want to make many requests to a speficic view or with specific dateranges, you can set date ranges for all future requests:

# Pass arguments to class init
c = GAClient(view_id="123456789", start_date="2019-09-01", end_date="2019-09-07") 
# or use methods to overwrite viewID or dateranges
c.set_view_id('123456789')
c.set_dateranges('2019-01-01', '2019-01-31')

request_body_1 = {
    'dimensions': {
        'ga:sourceMedium',
        'ga:date'
    },
    'metrics': {
        'ga:sessions'
    }
}

request_body_2 = {
    'dimensions': {
        'ga:deviceCategory',
        'ga:date'
    },
    'metrics': {
        'ga:sessions'
    }
}

response_1 = c.get_all_data(request_body_1)
response_2 = c.get_all_data(request_body_2)

Avoid sampling by taking data day-by-day

Important! Google Analytics reporting API has a limit of maximum 100 requests per 100 seconds. If you want to iterate over large period of days, you might consider adding time.sleep(1) at the end of the loop to avoid reaching this limit.

from datetime import date, timedelta
from time import sleep

import pandas as pd
from gaapi4py import GAClient

c = GAClient(view_id='123456789')

start_date = date(2019,7,1)
end_date = date(2019,7,14)

df_list = []
iter_date = start_date
while iter_date <= end_date:
    c.set_dateranges(iter_date, iter_date)
    response = c.get_all_data({
        'dimensions': {
            'ga:sourceMedium',
            'ga:deviceCategory'
        },
        'metrics': {
            'ga:sessions'
        }
    })
    df = response['data']
    df['date'] = iter_date
    df_list.append(response['data'])
    iter_date = iter_date + timedelta(days=1)
    time.sleep(1)

all_data = pd.concat(df_list, ignore_index=True)

Avoid "maximum 7 dimensions" restriction

If you store sessionId and/or hitId as custom dimensions (Example implementation on Simo Ahava's blog), you can circumvent restriction on maximum number of dimensions and metrics in one report. Example below:

If sampling starts to appear, try to break the set of dimensions into smaller parts and run queries on them.

one_day = date(2019,7,1)
c.set_dateranges(one_day, one_day)

SESSION_ID_CD_INDEX = '2'
HIT_ID_CD_INDEX = '5'

session_id = 'dimension' + SESSION_ID_CD_INDEX
hit_id = 'dimension' + HIT_ID_CD_INDEX


#Get session scope data
response_1 = c.get_all_data({
    'dimensions': {
        'ga:' + session_id,
        'ga:sourceMedium',
        'ga:campaign',
        'ga:keyword',
        'ga:adContent',
        'ga:userType',
        'ga:deviceCategory'
    },
    'metrics': {
        'ga:sessions'
    }
})

response2 = c.get_all_data({
    'dimensions': {
        'ga:' + session_id,
        'ga:landingPagePath',
        'ga:secondPagePath',
        'ga:exitPagePath',
        'ga:pageDepth',
        'ga:daysSinceLastSession',
        'ga:sessionCount'
    },
    'metrics': {
        'ga:hits',
        'ga:totalEvents',
        'ga:bounces',
        'ga:sessionDuration'
    }
})

all_data = response_1['data'].merge(response2['data'], on=session_id, how='left')

all_data.rename(index=str, columns={
    session_id: 'session_id'
}, inplace=True)

all_data.head()

# Get hit scope data
hits_response_1 = c.get_all_data({
    'dimensions': {
        'ga:' + session_id,
        'ga:' + hit_id,
        'ga:pagePath',
        'ga:previousPagePath',
        'ga:dateHourMinute'
    },
    'metrics': {
        'ga:hits',
        'ga:totalEvents',
        'ga:pageviews'
    }
})

hits_response_2 = c.get_all_data({
    'dimensions': {
        'ga:' + session_id,
        'ga:' + hit_id,
        'ga:eventCategory',
        'ga:eventAction',
        'ga:eventLabel'
    },
    'metrics': {
        'ga:totalEvents'
    }
})

all_hits_data = hits_response_1['data'].merge(hits_response_2['data'],
                                              on=[session_id, hit_id],
                                              how='left')
all_hits_data.rename(index=str, columns={
    session_id: 'session_id',
    hit_id: 'hit_id'
}, inplace=True)

all_hits_data.head()