fk

fk

HomeArchives

BI Visualization - Superset

#superset#BI28
AI Translation
This post is translated from Chinese into English through AI.View Original
AI-generated summary
The text discusses the use of Superset, an open-source tool for analyzing business data. The author explains that they previously used Elasticsearch and Grafana but found Grafana to be lacking in data mining and alert capabilities. They then installed Superset using Docker and set up their admin account. They imported data from Elasticsearch and explored various charting options. The author mentions that Superset has many advantages, such as support for Docker deployment, compatibility with various databases, and a wide range of charting options. However, they also note some limitations, including the lack of support for parameters in dashboards and the need to predefine queries for charting. Overall, the author concludes that Superset is a good tool for creating visually appealing dashboards, but it did not meet their specific needs.

Superset#

Description:#

Business needs require analyzing operational data. Previously, we used Elasticsearch + Grafana. However, Grafana, being a time-series visualization tool, was not very effective for data mining and alerting. So, we started looking for a new open-source tool; and that's where Superset comes in.

Installation:#

Chose to install using Docker
Searched hub.docker.com and used the official method:

$ docker run -d -p 8080:8088 --name superset apache/superset

With your local superset container already running...

  1. Setup your local admin account

    $ docker exec -it superset superset fab create-admin \
              --username admin \
              --firstname Superset \
              --lastname Admin \
              --email [email protected] \
              --password admin

  2. Migrate local DB to latest

    $ docker exec -it superset superset db upgrade

  3. Load Examples

    $ docker exec -it superset superset load_examples

  4. Setup roles

    $ docker exec -it superset superset init

  5. Login and take a look -- navigate to http://localhost:8080/login/ -- u/p: [admin/admin]

  6. Need to install an Elasticsearch plugin, as my database is Elasticsearch

	$ docker exec -it superset pip install elasticsearch-dbapi

Trying it out#

Importing a data source: Elasticsearch

After logging in --> go to the top right corner and click on "Setting" --> "Database Connections"

On the new page --> click on the top right corner "+DATABASE" to find Elasticsearch. Add it according to the format.
elasticsearch+http://192.168.100.68:9200/

The following steps are:

  • Create a Dataset; I understand it as a set of data extracted from the database based on a time format.
  • Create charts.
  • Add the charts to Dashboards.

Initially, the Elasticsearch data I imported did not have a timestamp and did not support sequential charts. The reason was that the date column in the imported data was recognized as a string format. After searching online, I found that I needed to format the date column as epoch_ms during the import process.

[Insert an image here]

However, my attempts were unsuccessful. In the end, I deleted the data source and added the Elasticsearch data source again.

Magically, the format recognized for the date column became Datetime. This solved all the problems.

In-depth exploration:#

Creating charts

1. Line, bar, and scatter plots#

  • These time series charts work fine.

  • The time range can be displayed with the chart. After adding it to Dashboards, the display can be adjusted within the selected time range.

  • Advanced options allow for time period segmentation. Options include sum, average, maximum, minimum, etc.

2. Pie charts#

  • They display normally and can only show data within the pre-selected time range.

  • Compared to line charts, they do not provide datetime selection.

  • Sunburst charts (a type of compound pie chart) can be selected, and they look good.

3. Other types of charts#

Other types of charts also work well. Most of them passed the tests.

Summary:#

Advantages:

  • Supports Docker deployment, which I prefer.
  • Supports various databases for import. Tested well with Elasticsearch.
  • Offers a wide range of chart types, allowing for the creation of stunning dashboards.

Disadvantages:

  • Dashboards do not support parameters. This means that time selection and conditional queries are not possible.
  • The process of creating charts is cumbersome, requiring the creation of a Dataset after importing data.
  • Charts can only be created based on pre-defined queries. It's quite rigid!

Overall, it's still a good tool, just not suitable for my needs. If you only need to analyze pre-defined data and create beautiful dashboards, it can be quite impressive!

Loading...
Ownership of this post data is guaranteed by blockchain and smart contracts to the creator alone.