Skip to main content Link Menu Expand (external link) Document Search Copy Copied

Fetch Data into a Jupyter Notebook

Time required: 5 minutes


Prerequisites

You must have already:

  • Signed up for a Matatika account
  • Created a workspace through the Matatika app or API
  • Published a dataset or access to an existing dataset

Introduction

Each dataset has a data endpoint, which returns live data from the database workspace schema based on the dataset query. The Matatika library fetch method is used to tap into this endpoint and return a snapshot of the dataset data. Using a Jupyter Notebook, we can create an interactive environment to fetch some data and perform transform and visualisation operations.

You can follow along with this guide using our simple_jupyter_fetch example notebook.


Fetching Data

Dataset data can be retrieved by invoking fetch as follows:

from matatika.library import MatatikaClient

# create the client and call 'fetch'
client = MatatikaClient(auth_token, endpoint_url, None)
data = client.fetch(dataset_id)

By default, the method will return a Python dictionary object constructed from the raw API response. From here, with the use of data-centric libraries such as pandas, NumPy or SciPy, it becomes easy to begin analysing, transforming and visualising the data in useful ways.


Using the Data

We can create a pandas.DataFrame using the from_dict method, and supplying the Matatika client-library fetch method return value as the argument.

import pandas as pd

# create the dataframe from the dataset data dictionary
df = pd.DataFrame.from_dict(data)
df.head()

The resulting dataframe can be visualised using the plot method, which functions as a wrapper for the plotting backend (by default this is Matplotlib).

df.plot()

total users plot

After some data clean-up, processing, and visualisation adjustments, it is possible to create plots that offer tailored insights.

total users plot with rolling average band