Pentaho Review

Pentaho BI Suite Review: Final Thoughts – Part 6 of 6


Introduction

This is the last of a six-part review of the Pentaho BI suite. In each part of the review, we will take a look at the components that make up the BI suite, according to how they would be used in the real world.

Data Mining

In this sixth part, originally, I'd like to at least touch on the only part of Pentaho BI Suite we have not talked about before: Data Mining. However as I gather my materials, I realized that Data Mining (along with its ilks: Machine Learning, Predictive Analysis, etc.) is too big of a topic to fit in the space that we have here. Even if I try, the usefulness would be limited at best since at the moment, while the result is being used to solve real-world problems, the usage of Data Mining tools is still exclusively within the realm of data scientists.

In addition, as of late I use Python more for working with datasets that requires a lot of munging, preparing, and cleaning. So as an extension to that, I ended using Pandas, SciKit Learning, and other Python-specific Data Mining libraries instead of Weka (which is basically what the Pentaho Data Mining tool is).

So for those who are new to Data Mining with Pentaho, here is a good place to start, an interview with Mark Hall who was one of the author of Weka who now works for Pentaho: https://www.floss4science.com/machine-learning-with-weka-mark-hall

The link above also has some links to where to find more information.

For those who are experienced data scientists, you probably already made up your mind on which tool suits your needs best and just like I went with Python libraries, you may or may not prefer the GUI approach like Weka.

New Release: Pentaho 5.0 CE

For the rest of this review, we will go over the new changes that comes with the highly anticipated release of the 5.0 CE version. Overall, there are a lot of improvements in various parts of the suite such as PDI and PRD, but we will focus on the BI Server itself, where the largest impact of the new release can be seen.

A New Repository System

In this new release, one of the biggest shock for existing users is the switch from file-based repository system to the new JCR-based one. JCR is a database-backed content repository system that was implemented by the Apache Foundation and code-named “Jackrabbit.”

The Good:

  • Better metadata management
  • No longer need to refresh the repository manually after publishing solutions
  • A much better UI for dealing with the solutions
  • API to access the solutions via the repository which opens up a lot of opportunities for custom applications

The Bad:

  • It's not as familiar or convenient as the old file-based system
  • Need to use a synchronizer plugin to version-control the solutions'

It remains to be seen if this switch will pay off for both the developers and the users in the long run. But it is stable and working for the most part, so I can't complain.

The Marketplace

One of the best feature of the Pentaho BI Server is its plugin-friendly architecture. In version 5.0 this architecture has been given a new face called the Marketplace:

This new interface serves two important functions:

  1. It allows admins to install and update plugins (almost all Pentaho CE tools are written as plugins) effortlessly
  2. It allows developers to publish their own plugins to the world

There are already several new plugins that is available with this new release, notably Pivot4J Analytics. An alternative to Saiku that shows a lot of promises to become a very useful tool to work with OLAP data. Another one that excites me is Sparkl with which you can create other custom plugins.

The Administration Console

The new version also brings about a new Administration Console where we manage Users and Roles:

No longer do we have to fire-off another server just to do this basic administrator task. In addition, you can manage the Mail server (no more wrangling configuration files).

The New Dashboard Editor

As we discussed in Part V of this review, the CDE is a very powerful dashboard editor. In version 5.0, the list of available Components are further lengthen by new ones. And the overall editor seems to be more responsive in this new release.

Usage experience: The improvements in the Dashboard editor is helping me to create dashboards for my clients that goes beyond the static ones. In fact, the one below (demo purposes only) has the interactivity level that rivals a web application or an electronic form:

NOTE: Nikon and Olympus are trademarks of Nikon Corporation and Olympus Group respectively.

Parting Thoughts

Even though the final product of a Data Warehouse of a BI system is a set of answers and forecasts, or dashboards and reports, it is easy to forget that without the tools that help us to consolidate, clean up, aggregate, and analyze the data, we will never get to the results we are aiming for.

As you can probably tell, I serve my clients with various tools that makes sense given their situation, but time and again, the Pentaho BI Suite (CE version especially) has risen to fulfill the needs. I have created Data Warehouses from scratch using Pentaho BI CE, pulling in data from various sources using the PDI, created OLAP cubes with the PSW, which ends up as the data source for the various dashboards (financial dashboards, inventory dashboards, marketing dashboards, etc.) and published reports created using the PRD.

Of course my familiarity with the tool helps, but I am also familiar with a lot of other BI tools beside Pentaho. And sometimes I do have to use other tools in preference to Pentaho because they suit the needs better.

But as I always mention to my clients, unless you have a good relationship with the vendor to avoid paying hundreds-of-thousands per year just to be able to use tools like IBM Cognos, Oracle BI, or SAP Business Objects, there is a good chance that the Pentaho (either EE or CE version) can do the same for less, even zero license cost in the case of CE.

Given the increased awareness on the value of data analysis in today's companies, these BI tools will continue to become more and more sophisticated and powerful. It is up to us business owners, consultants, and data analysis everywhere to develop the skills to harness the tool and crank out useful, accurate, and yes, easy-on-the-eyes decision-support systems. And I suspect that we will always see Pentaho as one of the viable options. A testament to the quality of the team working on it. The CE team in particular, it would be amiss not to acknowledge their efforts to improve and maintain a tool this complex using the Open Source paradigm.

So here we are, at the end of the sixth part. Writing this six-part review has been a blast. And I would like to give a shout out to the IT Central Station who has graciously hosted this review for all to benefit from. Thanks for reading.

Disclosure: I am a real user, and this review is based on my own experience and opinions.
7 visitors found this review helpful
Add a Comment
Guest
Sign Up with Email