DATA STORIES | DATA LITERACY | KNIME ANALYTICS PLATFORM

Teaching Business Analytics with KNIME at a Business School

Practices and experiences from the classroom at the University of New Brunswick

Authors: Dongmin Kim, Professor and Jong-Kyou Kim, Assistant Professor, University of New Brunswick

Photo by Priscilla Du Preez 🇨🇦 on Unsplash.

This article is a readapted version of a presentation at KMIS (The Korea Society of Management Information Systems) Spring Conference in 2024.

Teaching business analytics is increasingly prominent in university curricula due to its critical role in modern decision-making processes. As businesses generate vast amounts of data, the ability to analyze and derive actionable insights becomes essential for competitive advantage. This demand for data-driven decision-makers has led to a surge in business analytics programs, equipping students with skills in data analysis, statistical methods, and predictive modeling, preparing them to meet the evolving needs of the industry.

In this article, we would like to share our experiences using KNIME as a teaching tool for business analytics related topics. We have taught a couple of Management Information Systems (MIS) courses at a business school in 2024, targeting students with limited technical backgrounds.

We’ll first begin by introducing KNIME and its functions, followed by sharing examples of KNIME workflows used in the course. Next, we discuss initial student feedback and conclude pondering our future plans to advance the teaching of business analytics in academic settings.

Introduction to KNIME

KNIME Analytics Platform is a free and open-source visual programming-based tool used in business analytics. In KNIME, users create a workflow by dragging and dropping customizable nodes. Each node represents a set of programming instructions, making a workflow with no code or low codes. For instance, to conduct decision analysis in KNIME, users only need to use five nodes, as shown in figure 1:

1. Read the Data: Use a CSV Reader node to import data.
2. Split the Data: Divide the data into two groups, training data and test data, using a Partitioning node.
3. Build a Model: Develop a decision tree model using the training data with a Decision Tree Learner node.
4. Apply the Model: Use the Decision Tree Predictor node to make predictions using the test data.
5. Evaluate the Model: Assess the model’s performance using a Scorer node.

Figure 1. KNIME Workflow: Training and applying a Decision Tree.

Each node is configurable. For example, as shown in figure 2, when you configure the Partitioning node, you can specify the split percentage and training data selection methods, such as random selection or stratified sampling.

Figure 2. Configuring a Partitioning node.

As demonstrated in the examples above, KNIME allows users to focus on what tasks need to be done without needing detailed coding skills. This makes it easier for users to handle most business analytics tasks without learning a programming language, thus lowering the entry barrier to business analytics.

The main functions of KNIME include:

• Data cleaning, transformation, integration, and enrichment
• Data visualization
• Automating business analytics workflows
• Training machine learning models and making predictions
• Deploying models across an organization (requires a paid server version of KNIME)

Additionally, KNIME allows the integration of Python or R code into its workflows, enabling users to combine KNIME with Python or R to create more efficient workflows.

According to the official KNIME website, KNIME has over 300,000 users in more than 60 countries. KNIME offers a range of courses tailored to different data personas (see figure 3). Most of these online courses are available at no cost. KNIME also provides free books that users can refer to while working with the software.

Figure 3. KNIME Learning paths.

Most business managers in an organization use business analytics occasionally and usually don’t have any programming experience. However, with KNIME, they can still perform most analytics tasks without needing to learn a programming language.

On the KNIME Community Hub (https://hub.knime.com/), the free repository of shared knowledge of the KNIME community, there are 20,000+ workflows, 2000+ components, and 230+ extensions available that have been used in real workplaces. For instance, a manager can easily import a pre-made workflow for churn analysis (see figure 4). Simply by dragging the highlighted yellow icon to their local workflow editing screen, they can import the entire workflow and adapt it to suit their organization’s needs.

Figure 4. Workflow for Churn Analysis on the KNIME Community Hub.

Using KNIME in a MIS course

We introduced KNIME in Management Information Systems (MIS) courses during the winter of 2024. Approximately seven hours of class time, spread over four sessions, were dedicated to discussing KNIME.

During this time, students completed three key tasks using KNIME:

1. Transforming data into a format suitable for analysis, such as concatenating, joining, and unpivoting data (see figure 5 for a workflow example).
2. Analyzing data from a superstore (see figure 6).
3. Updating the Order and Customer Tables with transactions and returns received in December.

Figure 5. Concatenate two sheets into one.

Figure 6. A workflow to analyze sales of a superstore.

Surveying students on their KNIME experience

At the end of the winter 2024 term, students in an undergraduate Management Information Systems (MIS) course and an MBA MIS course were invited to participate in a D2L survey about KNIME. Students who participated voluntarily received a 0.5% — 1% course credit boost. Although the survey was not anonymous, the authors assured that all student-related information would be removed before analyzing the data.

The survey consisted of three open-ended questions:

1. What did you like about KNIME?
2. What did you dislike about KNIME?
3. Your thoughts about inclusion of KNIME in the MIS course for data cleaning and preparation.

Out of 101 students, 69 participated in the survey. We’ll report part of the preliminary results hereafter.

1. What did you like about KNIME?

Most students appreciated KNIME because it is relatively easy to use. Another significant advantage they noted is that it’s free of charge, allowing them to continue using it in the future.

Figure 7. What did you like about KNIME?

A few answers are reported below:

Table 1. What do you like about KNIME?

2. What did you dislike about KNIME?

Some students found KNIME overwhelming at first due to its numerous configuration tabs. They also mentioned that tasks seemed to take longer compared to using MS Excel. Additionally, a few students doubted they would use this tool in their future workplaces.

Figure 8. What did you dislike about KNIME?

A few answers are reported below:

Table 2. What do you dislike about KNIME?

3. Your thoughts about inclusion of KNIME in the MIS course for data cleaning and preparation

Most students were positive about including KNIME in the MIS course.

Figure 9. Inclusion of KNIME in MIS courses?

A few answers are reported below:

Discussion and Future Plan

Students in both undergraduate and MBA MIS courses generally had positive discussions about KNIME in class.

Most appreciated the tool for its ease of use and the fact that they could continue using it after the course. However, some found it overwhelming initially due to the many configuration tabs for each node. Another concern was that KNIME seemed to take longer for certain tasks compared to MS Excel, possibly because students didn’t fully grasp that KNIME workflows can be reused for repetitive tasks. To improve understanding and acceptance, future classes will discuss the benefits of tools like KNIME before diving into how to use them.

The authors plan to adopt KNIME in a Business Analytics course in the future.