DSensei

Introduction

dsensei-demo.mp4

The video is muted by default, 🎧 Unmute for audio explanations and improve your viewing experience!

Quick Start

Live Demo

https://app.dsensei.app

Running Locally

docker run -p 5001:5001 dsenseiapp/dsensei:latest

Open http://localhost:5001

Table of Contents

• What is DSensei
• Setup
• Contact

What is DSensei

DSensei is an AI-powered key driver analysis engine that can pinpoint the root cause of metric fluctuations within one minute. We save data teams hours to days of manual work on root cause analysis and help organizations uncover critical drivers and segments that are otherwise easy to overlook.

DSensei overcome the limitation of existing BI tools to empower you to understand the "why" behind metric fluctuations to inform better business decisions more effectively. Checkout our blog for more details.

Setup

There are multiple ways to run DSensei on your machine.

Using Docker (recommended)

The recommended way is to use the official Docker image. Make sure you have Docker installed on your system, then run the following command:

If you use CSV data source:

docker run -p 5001:5001 dsenseiapp/dsensei:latest

If you use BigQuery data source:

gcloud auth application-default login
docker run -p 5001:5001 -v ~/.config:/root/.config -e GCLOUD_PROJECT=$GOOGLE_PROJECT dsenseiapp/dsensei:latest

Replace the $GOOGLE_PROJECT with your own GCP project name.

This will pull the latest version of the DSensei-insight image and start the application on port 5001.

Running from code

To run the application locally without Docker, you need to have python3 and nodejs-18 installed on your system, then follow these steps.

1. First, navigate to the frontend directory and install dependencies using pnpm pnpm install or npm npm install
2. Then build the frontend npm run build
3. Switch to the backend directory and install python dependencies: python install -r requirements.txt
4. Finally, run the application by executing the following command in the backend directory flask run -p 5001

Contact

Please submit your bug report or feature request directly on github or in our discord group. We appreciate all your feedback!