I am quite a KNIME Fanboy, but seeing the recent developments on MS side makes me wonder where KNIME will stand in few years from now in comparison to big tech software.
Especially with the Data Marts integration in Power BI (still in beta, but soon to be released), there will be a powerfull tool enabling users in the ETL and data modelling regime. Data manipulation is done in an environment, which most business users are anyways familiar with, namely MS Office-like syntax and look&feel. Code integrations are possible as well. And the BI capabilities, well… Let’s phrase it like this: The typical use case in our organization is 1) use KNIME for data prep 2) push it to BI service and 3) visualize it in BI Service.
One thing that is currently not genuinely integrated in the Power Universe, is the ML part. Yet, in my experience at least, KNIME is mostly used by business users for automation tasks. ML is still practitioned mostly in native coding environments and I bet, it is just a matter of time, until big tech will add this in form of low coding to their portfolio, if it not already exists.
What is your impression and feeling about the future? Anyone here, who did an analysis/comparison/ or has actual experience working with the two KNIME and PowerBI + Datamarts?
It’s hard to compare the two. To me, this feels like comparing KNIME with Tableau Prep. Tableau Prep is a nifty tool for, lets say, simple but resource heavy calculations that I would rather just push off to Tableau’s servers. This way I can save KNIME Analytics resources for nodes that I need to run or are just easier/quicker to run locally. I would or could never have Tableau Prep take an entire ETL workflow it’s functionality is just nowhere as close to KNIME.
So I think it will be similar in PowerBI’s case - Data Marts would be secondary to PowerBI’s core functionality, which is analysis and visualization.
Does Tableau prep or the Addition of ETL capabilities into Power BI replace KNIME functionality for you? Perhaps, if that is all you are doing with it… Does it in any way compete with KNIME’s overall functionality and ability to incorporate into any platform or perform nearly any function? Absolutely not.
I actually see this trend moving in the other direction for much of the business world. Why work within the extreme platform limitations of Power BI and Tableau once we can achieve a similar fluid user interface in KNIME? I personally believe that the limitations of “visualizations” are quite frustrating Vs data apps that provide insights while also performing unlimited custom functionality based off of user interactions.
KNIME is moving in the direction of an ideal “open slate” that is only limited by the user’s imagination. Within a few years I believe that they will have the front end user interaction capabilities to build entire custom platforms solely within its infrastructure.
Greetings,
Working on Knime and Power BI simulataneously is very common practice in many industries. But,
I am seeking to deploy the “prediction whats if” or “model” into power bi in any form from knime, by exporting any table form. I know it is possible, But I am not sure how it will be possible excetly ? Please someone raise hand in this issue. So that we do not need to use heruko for any data science based project for any stakeholders or executives chief.
Future of the product usually not in the hands of enthusiast. It mostly depends on partners and development teams. If they continue to use KNIME as competitive product then there is a future. In other case it will be a product for students and enthusiasts.
in an enterprise environment the ETL decision is often not decided by the users but rather by the strategy of the company. And about the future - everything will be automated no matter the tool. Just my “2 cents”
Hi Daniel, Still now there is everything automated. Like rapidminer “Auto Data Cleasning” and “Auto ML”. That is not alternative of the coding and model optimisation or building or feature engineering. Without coding products are like “Packaged Food”. These still existing !!! I do not think that is a big issue. Chek the gartrner leader boards. In that case, Knime is not that much “Blind Therapy” product. I love knime not becuse of being “Open source”, I love it for extreme level customisation with or without coding/low-coding/connectivity.
@wisemanleo Power BI and Tableau Prep are in my opinion playing in different leagues, whilst tableau completely lost its edge over the recent years. Imho, tableau actually cannot justify its immense costs anymore vs the PowerBI. But that’s a different topic
@iCFO I am completely on your page. If it would be after me, I would choose to stay platform independent and providing the most functionality in one tool. Nevertheless, my experience also shows that “converting the mindset” of many business users is difficult, despite the potential for end-to-end automation and a crazy analytical toolbox in KNIME. Approximately 20% of colleagues adopt KNIME very well, some of them even become enthusiasts. The other 80% will stay in their Excel world, which is now enhanced by ETL (Datamarts & Powerapps) capabilities and fairly good visualizations (Power BI) + RPA functionalities (Power Automate) etc. etc… And since BI tools are anyways used, why then use another tool for data wrangling? In a worst case scenario, KNIME would stay in a niche then, where big tech platforms further evolve. This is also a question of how much capital KNIME will be able to generate to stay competitive. Not sure how that works with the current business model.
I guess the steps undertaken by KNIME go into the right direction though, as it seems for me that there are still some entry hurdles in terms of UX and ease of adoption, i.e.,
Improved UX and modern Interface. Easier entry point and appeal for users
Incorporating dynamic visualization and increasing the potential for data apps
Opening the platform for node development with python
Getting rid of the fat eclipse client and become more slim
“Thinking more corporate” with the KNIME Business Hub, since the KNIME Server in its current state brings the functionality one needs, but not more in my opinion…
Let’s see, I will put my hands on PowerBI in the next few weeks. Maybe I can share some further insights here.
I just found this discussion by accident but I completely agree with you @Residentstiefel .
I personally see KNIME become very popular with ETL use cases, like the “automate your Excel stuff”. When I speak to colleagues, I often try to convince them to do more or something different, e. g. text mining is a very attractive use case in finance departments (my niche so to say).
The argumentation goes like this: now that you know, what a node is, what a flow variable is and what this traffic light means, why don’t you take that knowledge to the next level?
I feel (at least a little bit) KNIME could do more in raising the awareness for the vast amount of people that come from the ETL point of view.
Looking at PowerBI (or the greater Power-Suite so to say): I think their biggest advantage - especially in business environments - is the integration part. People use Office, Excel etc anyways, so all the other tools feel quite familiar. The integration into MS Teams e. g. is a killer feature for me as Teams - as per Microsoft’s announcement - will be the core of their end user products in the future.
On the other hand, this is Microsoft … and this is only half meant as a joke. They have a longstanding history in missed opportunities.
Looking at PowerBI: I love the data visualization for everyone approach. But then, why would anyone, who does NOT have masochistic tendencies, learn DAX if they can use KNIME?
I started to do a lot of dashboards recently and I simply reject to stuff in DAX. But again, we’re reducing KNIME to ETL tasks as well.
What I would love to see are more stories of “we come from ETL and now we do this and that” (e. g. Text Mining, Predictions & Forecasting etc).
The challenge I see is that if a company goes “cloud” it often starts using the services from the big 3 cloudproviders as well and (whether I like them or not) can do the ETL part as well and are already “integrated”
