KNIME offers a delightful blend of power and ease-of-use. It uses a workflow (flowchart) interface. It offers you a collection of icons that represent steps in your analysis. You position each icon in a diagram and connect them with arrows that represent the flow of data or models from step to step. Double-clicking on an icon will open a dialog box allowing you to set the parameters of how that step should run.
You can document the workflow by clearly labeling each step, and by color coding entire sections of the diagram. As a result, you can look at another person's analysis and quickly get the big picture of what was done. If a diagram gets too complex, you can select a whole section of it and collapse it into a "metanode" so that one icon then represents all the steps. It's also easy to rerun an analysis on new data by simply changing the node(s) that read in data. Note that KNIME does not have a scripting language, nor does it need one.
I particularly like the company's view of open source software. The full desktop version of KNIME is free and open source, regardless of how much data you have. Its closest competitor price-wise is RapidMiner, which is free only if you have fewer than 10,000 cases. KNIME also offers excellent integration with a wide range of other open source software such as: Python, R, Spark, and even ImageJ for image analysis.
The server version of KNIME has a commercial license, i.e. not free. The company also sells software to let work groups share node libraries. For example, if I develop a new node in, say, R, and email it to everyone, then when I improve it, I'll have to email it again. Everyone using it would have to edit their work flows, but at least that's a free solution. However, if we purchase the collaboration extensions, I would make a change to the shared node, and everyone's work would be updated automatically. I appreciate having so much control over what to spend on a KNIME solution. Review collected by and hosted on G2.com.
For data mining or machine learning tasks, it's quite comprehensive, very similar to what RapidMiner, SAS Enterprise Miner, or SPSS Modeler can do. However, compared to a full statistics package such as R, SAS, SPSS, or Stata, it's lacking quite a few statistical methods.
You would have to lean on R or Python if you need statistical analysis rather than data mining. Each new version of KNIME adds more statistics, but it looks like it will be quite a while before they can offer full statistical capabilities. Unfortunately, no one tool offers comprehensive data mining, statistical modeling, and ease-of-use. In the ratings section, I'm giving it full marks as a data mining / machine learning tool. Review collected by and hosted on G2.com.
At G2, we prefer fresh reviews and we like to follow up with reviewers. They may not have updated their review text, but have updated their review.
The reviewer uploaded a screenshot or submitted the review in-app verifying them as current user.
Validated through a business email account
Organic review. This review was written entirely without invitation or incentive from G2, a seller, or an affiliate.
