I first discovered Knime while i was working on a project for Paris Dauphine University in 2015. I was looking for an open source tool that can help me to automate some ETL & reporting tasks for the library …. then i discovered the open projet Knime! It was really great! Knime allowed me to establish a reporting solution with the included Birt reporting engine! Used in conjunction with a french librarian’s tool called Ez-Paarse for parsing webpages! the task was quite challenging with millions of « anonymous » logs tracing the electronic ressources usage of the library! But Knime made the task simple and fun! Here is a screen capture of some of the workflows developped:
Screenshot of one workflow used to establish the reporting for the library (2015)
Each node was configured, and i connected the data with a local Mysql Database monthly updated.
Knime is very rich, and very intuitive and visually attractive compared to other classical tools! My first contact with Knime reminded me of some very good quality commercial solutions like IBM SPSS Modeler or SAS Entreprise Guide … but this time it was for free!🙂 Also Knime, is having a huge advantage : Extensions!
- Social media King (connect the api & play): This extensions can for example be connected with social networks like Twitter, Youtube, and also Google! You can then do some text mining for Tweets, scrapped webpages, or even Youtube videos for given keywords! This can be very practical for a marketing departement who is conducting monitoring activities on social media, or for consulting firms that want to track a brand or even collect detailed data.
- Forecast massive series with ease: I used to work on a forecasting project for Suez environment (energy) in Casablanca Morocco, the task was to find a good solution to conduct large automatic forecasting solution to predict the consumption for more than 2 millions consumers. I have consulted SAS & also SPSS and conducted some statistical and functional tests to observe how our statistical systems will perform with more than 120 millions of rows (5 years of historical data), some solutions were expensive or not completely appropriate for this task! Knime has the solution with the Arima extension & Time series! it needs some fine tuning, but the tool is perfect!
- Integration with R & Weka : Are you looking for more packages? Facebook? structural equations models?! Bayesian nodes?! don’t worry! you can make the tool integrated with R extensions and execute your code according to your project needs!
The community behind Knime is very active and nice! if you have any question you will find a helpful « Knimer » to help you! Datamining is very simple & logical especially with the available current tools! Marketers shouldn’t be afraid to invest and learn those new skills of great value! if used with a structured approach, this kind of tools can leverage huge advantages for the corporations! It’s all about finding the right mix between creative & scientific design to answer concrete facts!
Knime can be a great new friend for the curious minds!🙂
More about Knime & use cases can be found here : https://www.knime.org/