-
Notifications
You must be signed in to change notification settings - Fork 13
Home
Welcome to the mparticle-python-sdk wiki!
This SDK is a helper library for the mParticle server-to-server HTTP API, it exposes mParticle's schema as simple models and provides an HTTP client interface. This SDK is stateless and will only send the data that you populate, whereas our mobile SDKs will automatically collect app and device information, session events, install events, and maintain persistence. Read this wiki for a general overview and examples, and contact our customer support team to make sure you're feeding the platform with the right data to power your integrations.
All data sent via the SDK must be encapsulated in a Batch object. Each Batch is associated with a single user. Batch objects must be associated with an environment (production
or development
) to properly silo your testing and production data.
import mparticle
batch = mparticle.Batch()
batch.environment = 'development'
Most use-cases require that data be associated with a user identity, for example:
- If you're also sending data to mParticle via our mobile SDKs, set a customer ID both via the mobile SDKs and this SDK so that mParticle can correctly associate data with each user.
- Several marketing automation and audience integrations are powered by email.
identities = mparticle.UserIdentities()
identities.customerid = '123456'
identities.email = 'user@example.com'
batch.identities = identities
The DeviceInformation
object describes a mobile device that should be associated with this batch. Crucially, it exposes properties for device identities (Apple IDFA and Google Advertising ID) which are required for nearly all mParticle Audience integrations.
device_info = mparticle.DeviceInformation()
//set any IDs that you have for this user
device_info.ios_advertising_id = '07d2ebaa-e956-407e-a1e6-f05f871bf4e2'
device_info.android_advertising_id = 'a26f9736-c262-47ea-988b-0b0504cee874'
batch.device_info = device_info
The mParticle audience platform can be powered by only sending a combination of user attributes, used to describe segments of users, and device identities/user identities used to then target those users.
//arbitrary example allowing you to create a segment of users trial users
batch.user_attributes = {'Account type': 'trial', 'TrialEndDate':'2016-12-01'}
Events are central to many of mParticle's integrations; analytics integrations typically require events, and you can create mParticle Audiences based on the recency and frequency of different events. All events should be associated with a timestamp reflecting when they actually occurred, otherwise they will be assigned a timestamp when mParticle receives them.
App Events represent specific actions that a user has taken in your app. At minimum they require a name and a type, but can also be associate with a free-form dictionary of key/value pairs.
app_event = mparticle.AppEvent('Example', 'navigation')
app_event.timestamp_unix_ms = example_timestamp
batch.events = [app_event]
Many integrations support the notion of user-lifetime value or eCommerce transactions, and the LTV Increase event is a special-case of AppEvent that includes a monetary value. LTV Increase events are easy to use though much less descriptive than ProductAction events.
app_event = mparticle.AppEvent('User Value Update', 'transaction')
app_event.ltv_amount = 19.99
app_event.timestamp_unix_ms = example_timestamp
batch.events = [app_event]
The CommerceEvent is central to mParticle’s eCommerce measurement. CommerceEvents can contain many data points but it's important to understand that there are 3 core variations:
- Product-based: Used to measure measured datapoints associated with one or more products
- Promotion-base: Used to measure datapoints associated with internal promotions or campaigns
- Impression-based: Used to measure interactions with impressions of products and product-listings
product = mparticle.Product()
product.name = 'Example Product'
product.id = 'sample-sku'
product.price = 19.99
product_action = mparticle.ProductAction('purchase')
product_action.products = [product]
product_action.tax_amount = 1.50
product_action.total_amount = 21.49
commerce_event = mparticle.CommerceEvent(product_action)
commerce_event.timestamp_unixtime_ms = example_timestamp
batch.events = [commerce_event]
The SessionStartEvent and SessionEndEvent should be used to describe the details of user session such as its length, which is a common metric used in many mParticle integrations. Additonally, length, recency, and frequency of sessions are powerful data-points by which an mParticle audience can be defined.
session_start = mparticle.SessionStartEvent()
session_start.session_id = 12345678
session_start.timestamp_unixtime_ms = example_timestamp
session_end = mparticle.SessionEndEvent()
session_end.session_id = session_start.session_id # its mandatory that these match
session_end.session_duration_ms = example_duration
session_end.timestamp_unixtime_ms = example_timestamp + example_duration
batch.events = [session_start, session_end]
Use these events to represent the nuances of intra-session user-behavior with your app. Install events are crucial to power many attribution integrations and to judge the success of marketing campaigns.
install = mparticle.ApplicationStateTransitionEvent.get_install_event()
install.timestamp_unixtime_ms = example_timestamp
upgrade = mparticle.ApplicationStateTransitionEvent.get_upgrade_event()
upgrade.timestamp_unixtime_ms = example_timestamp
foreground = mparticle.ApplicationStateTransitionEvent.get_foreground_event()
foreground.timestamp_unixtime_ms = example_timestamp
background = mparticle.ApplicationStateTransitionEvent.get_background_event()
background.timestamp_unixtime_ms = example_timestamp