Skip to content

Quickstart Guide

We will install VerdictDB, create sample, and issue a simple query to VerdictDB. In this Quickstart Guide, we will use MySQL for VerdictDB's backend database. See Connecting to Databases for the examples of connecting to other databases.



Create an empty Maven project and place the verdictdb dependency in the <dependencies> of your pom.xml.


pyverdict is distributed with PyPI. No installation of VerdictDB is required.

To insert data into MySQL in Python without pyverdict, we could use pymysql.

Note: Prerequisites

pyverdict requires miniconda for Python 3.7, which can be installed for local users (i.e., without sudo access).


<!-- To use MySQL, add the following entry as well: -->
pip install pyverdict
# use the following line for upgrading:
# pip install pyverdict --upgrade

# install pymysql to use MySQL
pip install pymysql

Create Sample

Assume a table myschema.sales already exists. Create a connection to VerdictDB. Create a special table called a "scramble", which is the replica of schema.sales with extra information VerdictDB uses for speeding up query processing.

Connection verdict =
    DriverManager.getConnection("jdbc:verdict:mysql://localhost", "root", "");
Statement vstmt = verdict.createStatement();
vstmt.execute("CREATE SCRAMBLE myschema.sales_scrambled from myschema.sales");
verdict_conn = pyverdict.mysql(
verdict_conn.sql('CREATE SCRAMBLE myschema.sales_scrambled from myschema.sales')

Run Queries

Run a regular query to the scrambled table to obtain approximated results. In PyVerdict, The query result is stored in a pandas DataFrame. The values may vary.

ResultSet rs = vstmt.executeQuery(
    "SELECT product, AVG(price) "+
    "FROM myschema.sales_scrambled " +
    "GROUP BY product " +
    "ORDER BY product");
df = verdict_conn.sql(
    "SELECT product, AVG(price) " +
    "FROM myschema.sales_scrambled " +
    "GROUP BY product " +
    "ORDER BY product")