It's quite normal to find Industrial Engineers in all kinds of different career paths but that doesn't mean the entry into a different field always runs seamlessly. Looking back, there is one thing in particular I wish I had known before starting my career in Business Intelligence (BI) four years ago: The importance of Machine Learning (ML). Here's why.
I am an Industrial Engineer and chose to pursue a career in BI. This, on its own, is not unusual - we Industrial Engineers are known to be jack of all trades. We are trained to understand and improve processes across the entire engineering spectrum. That's why it's not strange to find Industrial Engineers scattered throughout many occupations with other job titles and why having a broad understanding is more beneficial to us than extensive understanding in a single domain such as BI.
Levelling up for Business Intelligence
The main building blocks I use for BI from my Industrial Engineering background are the following two:
- Operations Research which concerns decision support and optimisation with mathematical models and normally entails large amounts of data.
- Information Systems which involves the arrangement of above mentioned data, the processes and the technology that interact with people to support and improve this decision-making and optimisation exercise.
These are best understood with an example from the process of a project I completed that had to do with restocking ice cream vendors. On a high level, the overall process is this: An ice cream counter goes to ice cream shops to count the supplier's ice creams, after which replenishment ice cream is sent from the supplier. A more granular process with references to Operations Research and Information Systems on the other hand would look as follows:
- The ice cream counter uses an application on his mobile phone (the technology) to capture the total number of ice creams left at the ice cream shop.
- The total number of ice creams left (the data) is fed back to the supplier.
- The supplier uses a set of rules, logic and calculations to determine which and how many ice creams need to be replenished and when (the decision support and optimisation).
In BI, I work with processes like this one to help businesses understand and deploy data driven projects in order to add significant decision making power. Normally, I provide the following statement from Forbes when people ask about my responsibilities:
"A Business Intelligence Analyst (BIA) is a professional, either a consultant or full-time employee, that uses data to inform their recommendations to companies and help guide decisions based on the market or trends. Heavily reliant on trends, business intelligence analysts are always combing through data to identify new opportunities for companies and articulating the business meaning of the data results to stakeholders."
While in and of themselves, my Industrial Engineering building blocks are incredibly valuable, I realised very soon that I needed to level up on certain other skills to successfully live up to this job description. The main areas I needed to upskill were:
- Data Visualisation and BI Platform Experience, because it would allow me to tell stories with the data in a tool with high user experience - this increases the accessibility of the data which increases understanding and buy-in, all of which are key for successful BI projects.
- Machine Learning (ML), because automation from advanced mathematical models enhances insights from BI projects.
- Coding, because BI professionals can master up to 17 programming languages and I was nowhere near that number.
While we had touched on aspects of the above during my Industrial Engineering studies, I realised during my BI work that I needed to deep dive on these three skill sets.
The value of Machine Learning
About a year and a half ago, I identified the biggest area of improvement within my skill set as that of ML. I felt a real need to understand the concepts and how they can be applied in the work I do.
Take my restocking of ice cream vendors project: I didn't think that the model for determining the optimal replenishment ice cream was complex enough and I wanted something more sophisticated. What if I could use data to build an ML model that would automatically determine the replenishment ice cream amounts depending on the time of year? Furthermore, imagine this model adjusting the replenishment ice cream amounts as time goes by to stay relevant to variants such as the economy, demographic composition and weather!
The opportunities were endless. That's why I decided to teach myself ML through a few online courses. This has really helped me, because it:
- Enhances my analysis capabilities due to the bunch of new algorithms which have been added to my skill set.
- Develops my coding abilities because it forces me to use programming languages such as R and Python.
- Resonates with industry trends which provides the ease of mind that I'm spending my time on something that will last.
For me, that last statement is the most important. Why? Because it speaks for my assumption that ML is important for Industrial Engineers. So far, I've received three very interesting confirmations:
First of all, the Industrial Engineering faculty I graduated from posted an advertisement for an Associate Professor in Data Analytics and Machine Learning Applications in Industrial Engineering one year after I started my ML journey. They've clearly also realised that ML is a crucial skill for Industrial Engineers.
Secondly, the Executive Dean: Faculty of Engineering from North West University made some predictions about Industry 4.0 after which a few responded that the question can be answered with ML by removing it from its current position. See how closely it relates to Industrial Engineering and Operations Research?
Finally, the Eckerson Group who are consultants on the BI industry predict that the next stage of BI will be infused with ML, with some BI platforms already reflecting aspects thereof with advanced clustering algorithms, neural network forecasting and more. Instead of starting with a hypothesis and querying data to prove or disprove a hypothesis, BI professionals will expand and validate the insight gained from ML algorithms. This is why having a clear understanding of the logic behind ML algorithms is beneficial.
Resources
- Udemy's Machine Learning A-Zâ„¢ - Hands-On Python & R In Data Science: In-depth ML algorithm programming experience.
- University of Helsinki's Elements of AI: Very high level and easy to digest course on the topic (among other) of ML.
- Daniel Kunin's (from Brown University) Seeing Theory: Interactive course on statistics really gives a new perspective on statistics (especially those used in ML).
- D3.js: Jaw-dropping web based visualisations.
- Pythian's 4-part blog series on The Changing Role of BI Professionals: An industry leader's take on BI with a refreshing conclusion in a series of articles.
- Brandon Rohrer's (from Facebook) How Optimization for Machine Learning Works: A data scientist and influencer's take on ML in an easy to follow series.
- R2D3's A Visual Introduction to Machine Learning: A striking visual way to learn the basic principles of ML with specific focus on decision trees.
- SuperDataScience Podcast: The creators of Udemy's course on ML.
- Secrets of Data Analytics Leaders Podcast: Significant opinions from industry experts.
Anine is a data driven Industrial Engineer in the world of Business Intelligence. She is on a journey to become the best Business Intelligence professional she can be.
