Datagit

Datagit is a git based metric store library

>>> from datagit import github_connector
>>> from github import Github

>>> dataframe = bigquery.Client().query(query).to_dataframe()
{"unique_key": ['2022-01-01_FR', '2022-01-01_GB'...
>>> github_connector.store_metric(ghClient=Github("Token"), dataframe=dataframe, filepath="Samox/datagit/data/act_metrics_finance/mrr.csv", assignees=["Samox"])
'🎉 data/act_metrics_finance/mrr.csv Successfully stored!'
'💩 Historical data change detected, Samox was assigned to it'

Purpose

Non-moving data is a journey, in reality, the data moves, or drifts. The purpose of this library is

• to parse, sort, sanitize a metric dataset
• to convert it to CSV
• then to store it in a Github repository with clean commits for new data, or drifting data.

Getting Started

To get started with Datagit, follow these steps:

1. Create a new repository on GitHub called datagit (or whatever other name you prefer) with a README file.
2. Generate a personal access token on GitHub that has access to the datagit repository. You can do this by going to your GitHub settings, selecting "Developer settings", and then "Personal access tokens". Click "Generate new token" and give it the necessary permissions (content and pull requests).
3. In your data pipelines, when relevant, call store_metric with the following parameters
   • a github client with your token Github("Token")
   • your metric in a dataframe format
   • the path of metric in a with a csv format: "your_orga/your_repo/path/to/your.csv"
   • The owner of the metric

For instance

>>> from datagit import github_connector
>>> github_connector.store_metric(ghClient=Github("Token"), dataframe=dataframe, filepath="Samox/datagit/data/act_metrics_finance/mrr.csv", assignee=["Samox"])

That's it! With these steps, you can start using Datagit to store and track your metrics over time.

Example

>>> githubToken = "github_pat****"
>>> githubRepo = "ReplaceOrgaName/ReplaceRepoName"
>>> import pandas as pd
>>> from datetime import datetime
>>> dataframe = pd.DataFrame({'unique_key': ['a', 'b', 'c'], 'date': [datetime(2023,9,1), datetime(2023,9,1), datetime(2023,9,1)], 'amount': [1001, 1002, 1003], 'is_active': [True, False, True]})
>>> from github import Github
>>> from datagit import github_connector
>>> github_connector.store_metric(ghClient=Github(githubToken), dataframe=dataframe, filepath=githubRepo+"data/act_metrics_finance/mrr.csv")

Dataset

Datagit is base on the standard dataframe format from Pandas. One can use any library to get the data as long as the format fits the following requirements:

1. The first column of the dataframe must be unique_key
2. The first columns must have only unique keys
3. The second column must be a date

The granularity of the dataframe depends on every use case:

• it can be at very low level (like transaction) or aggregated (like a metric)
• it can contain all the dimension, or none

1st column: Unique key

The unique_key is used to detect a modification in historical data

In case you have duplicated lines, datagit will automatically rename them with -duplicate-n

  unique_key  value
0          A     10
1          B     20
2          C     30
3          B     40
4          C     50
5          C     60
6          D     70

         unique_key  value
0                A     10
1                B     20
2                C     30
3    B-duplicate-1     40
4    C-duplicate-1     50
5    C-duplicate-2     60
6                D     70

2nd column: Date

The date key is used to detect new historical data, or deleted historical data

Query Builder

Datagit provides a simple query builder to store a table:

>>> from datagit import query_builder
>>> query = query_builder.build_query(table_id="my_table", unique_key_columns=["organisation_id", "date_month"], date="date_month")
'SELECT CONCAT(organisation_id, '__', date_month) AS unique_key, date_month as date, * FROM my_table WHERE TRUE ORDER BY 1'

More examples here

Large Dataset

Partitionning

In case of more than 1M rows, partitionning is recomanded using the partition_and_store_table function.

>>> from datagit import github_connector

>>> very_large_dataframe = bigquery.Client().query(query).to_dataframe()
{"unique_key": ['2022-01-01_FR', '2022-01-01_GB'...
>>> github_connector.partition_and_store_table(ghClient=Github("Token"), dataframe=very_large_dataframe, filepath="Samox/datagit/data/act_metrics_finance/mrr.csv")
'🎁 Partitionning data/act_metrics_finance/mrr.csv...'

Drift

A drift is a modification of historical data. It can be a modification, addition or deletion in a table that is supposed to be "non-moving data".

Drift evaluator

When a drift is detected, the default behaviour is to trigger an alert and prompt the user to explain the drift before merging it to the dataset. But a custom function can be used to decide weather an alert should be triggered, or if the drift should be merged automatically.

Default drift evaluator

The default drift evaluator will open a pull request with a message containing the number of addition, modifications and deletions of the drift.

Custom drift evaluator

You can provide a custom evaluator which is a function with the following properties:

• parameters:
  • `data_drift_context``: a dictionnary with:
  • computed_dataframe (the metric up to date)
  • reported_dataframe (the metric already reported)
• return value:
  • A dictionnary containing:
  • "should_alert": Boolean, If True a pull request will be opened, If False the drift will be merged
  • "message": str, the message to display in the pull request, or the body message of the drift commit

No alert drift evaluator

In case you just want to store the metric in a git branch, this drift evaluator merge the drift in the reported branch without any alert.