top of page

Sparkflows on Incorta

Sparkflows-Incorta Integration Overview

The native integration between Incorta and Sparkflows unifies the strengths of the two platforms and unlocks the predictive power of Operational Lakehouse enabling Customers to:​

Build AI/ML-powered advanced analytical solutions without leaving Incorta

​

EnvironmentImplement AI/ML-powered horizontal (ERP Finance) and Industry vertical (Retail, CPG, Telco and BFSI)

 

Data Applications/Analytical Solutions Enhance

Incorta Data Applications with Predictive Analytics

End-To-End Application Integration

Platform Integration

Screenshot 2024-12-18 111955.png

Installation Requirements

Chidori - Compute connection

Navigate to Administration -> Configuration -> Global Connection.

​

Create a connection with the Chidori Connection.

Chidori is a technology as well as a product service provided by Incorta to enable Incorta to run Spark job without affecting the regular Incorta


Select Add Connection in the Connections section of the Administration page of Sparkflows

Create a Chidori Connection

Key Points
​

  • Specify the unique token for this connection

  • Specify the Postback URL where the Spark Job outputs will be posted back to Sparkflows UI

  • Specify the URL of Incorta Spark Cluster

  • Specify the main WorkflowExecutor CLass

  • Specify the location of the sparkflows fat jar

Create Incorta Storage Connection

Navigate to Administration -> Configuration -> Global Connection.
Create a connection with the Incorta Object Store.


Key Points

  • Specify the username and password of the Database

  • Specify the Database URL

  • Specify the database driver class (Incorta uses postgress)

  • Specify the hostname and API Key

Configure MLOps

Create a Dataset with Incorta

View Schema info on Dataset

Read from Incorta

Write to Incorta 

Design Data Preparation Workflow

Execute Workflow using Chidori Connection

Create MV in Incorta from Sparkflows Workflow

Generate Code from Sparkflows

Design ML Workflow

unnamed (7).png

Save ML Model

values (1).png

Incorta AI application for Retail - Customer Lifetime Value

Business Problem

 

  • How much to spend on acquiring a new customer ?

  • What are the Top 10 and Bottom 10 predictions of CLTV ?

  • What are the order and revenue distributions for the customers with low CLTV ?

  • Which segments have much higher CLV than others ?

  • How much a customer base is worth when valuing a company for purchase ?

  • How much does one need to invest in a current customer, i.e. spending on retaining, serving, & cross-selling ?

  • What are the most important touch points where the customers create value ?

Business Problem

Descriptive Analytics
​

  • Monthly Total Sales

  • Monthly Transaction Count

  • Monthly Sales Volume

  • Customer Store Visit Distribution

  • Country wise Invoice Count

  • Country wise Distinct Customer Count

  • Customer Transactions Distribution

Predictive Analytics
​

  • Comparative Analysis of Predicted CLTV and Actual Value

  • Top 10 Customers by predicted CLTV

  • Top 2 Customer Clusters by predicted CLTV

  • Bottom 2 Customer Clusters by predicted CLTV

  • Sales Revenue by Customers with High predicted CLTV

  • RFM Analysis of Customers

bottom of page