Experiments demo

Experiments demo.ipynb

Experiments demo.py

# coding: utf-8

# # Accessing pref-flipping experiments data
# 
# This notebook briefly demonstrates how to access the raw data for participants in experiments. 
# 
# You may be able to get the information you need from the [Experiments Viewer](https://moz-experiments-viewer.herokuapp.com/) dashboard.
# 
# If not, then continue reading for a quick introduction.

# In[1]:

from moztelemetry.dataset import Dataset


# In[3]:

cohorts = Dataset.from_source("telemetry-cohorts")
cohorts.schema


# Fetch some example data for a single day and a single experiment.

# In[9]:

experiment = "shield-public-infobar-display-bug1368141"
pings = cohorts.where(submissionDate="20170701")                .where(experimentId=experiment)                .records(sc)


# Cache the dataset since it's small and we'll be poking around in it several times.

# In[10]:

pings.cache()


# How many pings did we observe in total?

# In[11]:

pings.count()


# What were the different ping types observed for this experiment on this day?

# In[13]:

pings.map(lambda f: f["meta"].get("docType", "None")).countByValue()


# How many different `clientId`s submitted data for this experiment on this day?

# In[14]:

pings.map(lambda f: f["meta"].get("clientId", "None")).distinct().count()


# ### Where to go from here?
# 
# You will most likely want to fetch data for *all* days of an experiment, which you can do by leaving out the `submissionDate` parameter.
# 
# You will probably also want to fetch particular document types, so you can add a new `docType` filter.
# 
# You may also want to look at the `experiments_aggregates` data set which contains pre-aggregated results for each experiment and branch.