Reading a MongoDB collection into a pandas data frame (aggregation query)

You can use an aggregation query to filter/transform data in MongoDB before fetching them into a data frame. This allows you to delegate the slow operation to MongoDB.

Reading a collection from MongoDB into a pandas DataFrame by using an aggregation query:

import pdmongo as pdm
import pandas as pd

# First generate some data and write them to MongoDB
df = pd.DataFrame({'A': [1, 2], 'B': [3, 4]})
df.to_mongo(df, 'MyCollection', "mongodb://localhost:27017/mydb")

# Filter with an aggregate query and parse results into a data frame.
query = [{"$match": {'A': 1} }]
df = pdm.read_mongo("MyCollection", query, "mongodb://localhost:27017/mydb")
print(df) # Only values where A > 1 is returned

The query accepts the same arguments as the aggregate method of pymongo package.

Write MongoDB to a PostgreSQL table

You can write a MongoDB collection to a PostgreSQL table:

import numpy as np
import pandas as pd
import pdmongo as pdm
from sqlalchemy import create_engine

# Generate some data and write them to MongoDB
df = pd.DataFrame({'A': [1, 2, 3]})
df.to_mongo("MyCollection", "mongodb://localhost:27017/mydb")

# Read data from MongoDB and write them to PostgreSQL
new_df = pdm.read_mongo("MyCollection", [], "mongodb://localhost:27017/mydb")
engine = create_engine('postgres://postgres:postgres@localhost:5432', echo=False)
new_df[["A"]].to_sql("APostgresTable", engine)

Plot data retrieved from a MongoDB Collection

You can plot a collection retrieved from MongoDB

import numpy as np
import pandas as pd
import pdmongo as pdm
import matplotlib.pyplot as plt

# Generate data and write them to MongoDB
df = pd.DataFrame({'Value': np.random.randn(1000)})
df.to_mongo('TimeSeries', 'mongodb://localhost:27017/mydb')

# Read collection from MongoDB and plot data
new_df = pdm.read_mongo("TimeSeries", [], "mongodb://localhost:27017/mydb")
new_df.plot()
plt.show()

Changelog