How to Install KNIME

You can download the KNIME Analytics Platform software for your preferred operating system at the link below.

Link: www.knime.com/downloads

For Windows and macOS, you need to run the installation program and confirm the recommended settings. With Linux, you only need to unpack the archive (in your home directory).

Display with Linux Wayland: When you start KNIME, you’ll be notified that you should use Xorg for an optimal display, but you can still try Wayland. If the display isn’t satisfactory, you can log out and switch to Xorg in the login window.

After starting the program, a window opens with the instruction to select the workspace in which your programs and configurations are saved (see figure below). Check the Use this as the default and do not ask again box, and click Launch.

Once you’ve set or confirmed the workspace, the KNIME Workbench appears.

New GUI for Version 5.1: If you download the current version of KNIME, the user interface (UI) is structured differently by default than shown in the screenshots in this book, but that isn’t a problem at all. You only need to make one setting to be able to follow the instructions in this book: on the start page in the first figure below, click on the icon at the top right, and select Switch to KNIME classic user interface (see second figure).

You should get an overview of the structure of the software. If you’ve never dealt with development environments before, the UI may appear somewhat complex to you (see figure below). Don’t worry, soon you’ll be able to routinely create programs and make settings.

Following are areas of the KNIME Workbench:

1. Menu bar: Make central settings, search for and install updates, and so on.
2. KNIME Explorer: Access the project structure.
3. Workflow Coach: Activate this feature by clicking on it, and suitable nodes (e.g., graphic elements or modules) for your program will be recommended. This saves you a lot of time later during development, so activating is recommended.
4. Node Repository: See an overview of all existing nodes.
5. Outline: See a complete overview of the workspace. This can be interesting if you have large programs.
6. Console and Node Monitor: Find important issues of interest to you.
7. Description and KNIME Hub Search: Find information on selected nodes in the node description. Access and use ready-made programs (workflows) from the community in the hub search, for example.
8. Workflow Editor: Access the workspace in which the actual programming work takes place.

Configuration

Next, let’s install other required packages by going to Help > Install New Software in the menu bar. Under Work with, you can set the search to be performed on all pages. Then, search for Python, and select the items shown in the next figure.

• KNIME Conda Integration
• KNIME Python Integration
• KNIME Python Scripting extension

Click Next, accept the license conditions, and confirm with Finish. The selected packages will then be downloaded and installed.

Install the following packages in the same way:

• KNIME Deep Learning—Keras Integration
• KNIME Deep Learning—TensorFlow Integration
• KNIME Deep Learning—TensorFlow2 Integration
• KNIME Text Processing
• KNIME Image Processing

We still need to make some settings for Python, Keras, and TensorFlow. To do this, go to File > Preferences > KNIME > Python via the menu bar. Select Conda and wait briefly until KNIME has collected all the information about the installation. You’ll likely see a warning that some packages are missing. For this reason, we’ll now create an environment that contains all the required packages. To do this, click on New environment, and keep the suggested name for the environment (here py3_knime), as shown in this figure.

To ensure that programs with ANNs can be created later, you must create another environment for this by going back to File > Preferences > KNIME > Python Deep Learning via the menu bar. Then, select Use special Deep Learning as defined below, and choose Keras and Conda. You’ll then have to wait again until KNIME has collected information about the Python installation. Now click on New environment under Keras, accept the suggested name, and click on Create new CPU environment. Close the menu (next figure) by clicking on Apply and Close.

Create an environment for TensorFlow 2 in the same way. This prepares the development environment for visual programming.

Troubleshooting: Unfortunately, creating environments via KNIME doesn’t always work smoothly and depends on the versions of the operating system, Anaconda Navigator, Python, and KNIME. If you receive error messages, try updating Anaconda first. If the problem occurs when you create the environment for Python, you can select an older Python version (e.g., 3.9) from the dropdown menu before you select Create New Environment.

Test

We’ll test the installation with an Excel file, which you can create using Microsoft Excel or LibreOffice Calc, for example.

In KNIME Explorer, right-click on LOCAL, select New Workflow Group from the context menu, name it “Test”, and click Finish. Then, create a workflow named “Test-1” in this workflow group. After that, you need drag and drop the XLSX file into the workflow group (this is a folder) to copy it there. The folder structure should look like this.

Now, you can drag and drop the XLSX file into the workflow editor again. A creation wizard appears, as shown in the next figure, which usually recognizes all settings correctly. All you have to do now is confirm by clicking OK.

The Excel Reader module or node appears with a yellow dot. You can execute the module by right-clicking and then clicking Execute. The dot changes to green, and the content of the file gets displayed in the output (see next figure). To reset the module, you need to right-click and then select Reset.

Your development machine is now configured, tested, and ready for use. Now you can start developing AI models with KNIME.

Editor’s note: This post has been adapted from a section of the book Developing AI Applications: An Introduction by Metin Karatas.