Who needs 100% accuracy when 90% is more than enough? More often than not, practical, on the ground applications, that become very popular in use, depend on one simple algorithm that beats all complicated ones developed by scientists to impress other scientists. This is the story of someone who happens to be at the intersection of two important fields: engineering and data science.  Let me tell you something, if I want to develop a complicated neural network that does the desired task with an added accuracy/efficiency of 5% I would drop the idea before I even start. Let me give you an example.  Designing the fire alarm system is one of the most tedious tasks in engineering design. I heard so many ideas on how to automate the process by taking the dimensions of each room and make sense out of them, or at times, using machine learning to perform the task by gathering data from previous projects and so on.  While these ideas make sense, they actually don't (for the time being).  Sou

  Sources: https://isaacnewtonassignmentufcc.weebly.com/newton-and-the-apple.html As I was reading my Deep Learning book for my university course the other day, going through all the equations of how a linear regression machine learning algorithm reaches the parameters of a mathematical equation at 7:30am before work, with a good cup of Greek/Turkish coffee, a lightbulb went on in my head.  Let's rewind a bit back, for the past 2 weeks, I was already working on a machine learning algorithm to model the operation of an Air Handling Unit at work, and I was looking at complex mathematical/engineering equations which were supposed to be used to try and model the AHU operation.  So, that's when it clicked; why use very complex models with endless parameters, where real life is even much more complex to be modeled with equations from scratch, when in fact real-time data carries with it the complexity of the operations which, with a simple linear regression model and a few parameters,

While working on modeling the behavior of a component of the HVAC system using machine learning algorithms, I observed that when the data was changed to observe how the system would react, there were sharp movements in the model that didn't quite map out to reality.  Source: https://mindsports.nl/index.php/moving-forward-looking-back The issue bugged me. I knew there would be a solution.  One day, while I was having my morning coffee and reading a paper on path dependence in economics, I saw the solution in front of my eyes: to perfectly capture the behavior of the system, not only do I have to feed the algorithm the present data, but the previous (t-1) data as well!  Rushing to my laptop, I fixed the code to account for previous data as well as present data, and voila! It worked! I could easily replicate the behavior of the system with this very simple step. Often times, data scientists work with data they are not familiar with, which leads to over-complications in the code/algori

  Last month, we had one of the biggest projects that I have personally worked on. The scale was massive, and hence, the number of devices was enormous. If parametric/automation design was not an option, it would surely be a nightmare.  In the past, I had developed a code that would help engineers specify the wiring path for fire alarm devices, that basically included both, manual and automated efforts to save time.  But in this project, this was no longer applicable, the scale required full automation to deliver on time.  What to do? The first idea I had was to go back to math! I knew that there would be a mathematical solution to my problem, and my research led me to the Traveling Salesman Problem. A fascinating mathematical problem that was solved and even translated into python code.  Thankfully, Dynamo in Revit enables the use of Python, so I used the traveling salesman algorithm to define the optimum path for wiring the devices for each floor automatically. In 5 mins, thousands o

  Emperor Constantine I, Presenting the first Clash Free BIM Model of the City to the Blessed Virgin Mary We all have role models in our lives, people we look up to and wish we can be like (mine is Nassim Nicholas Taleb for example).  But can BIM Revit models have role models too? Yes they can! I'm pretty sure any design engineer who works with Revit has come across projects where the project is divided into several packages, with the client begging the consultant to have consistency across all packages, to which the consultant replies with "Sure!" but usually doesn't achieve the desired consistency due to human error. But, what if we can solve this problem with Machine Learning? Using Memory Based Reasoning algorithms and Decision Trees, I was able to achieve this just last week.  Two weeks ago, I was assigned to the 4th package of a project. The first three packages were already delivered to the client. I had two choices; one, go through all the models, find the pat

I’m about to finish my first year in the data science program at the University of Nicosia (part-time) and my 6th year as a telecom engineer at Dar (full-time). During the past year, I have seen things that totally shifted my career focus and I want to share my top takes on the topic and the future.  Top-down vs Bottom-up in corporations: Applying data science/AI/ML at work as an engineer and a data scientist in the making is vastly more different from hiring a pure data scientist to help engineers with their work. I see companies trying to catch the trend of ML/AI by hiring data scientists who don’t know anything about the engineering job and who start making “sophisticated” applications that really have minimal value and is mainly for show off. I have been creating ML/SQL/deep learning tools at work with very minimal resources and even minimal coding because I know how the output should be, what is the data and how the job works, and how others can use these tools after me. Some were

This is my first attempt to analyze (visualize) COVID data available online (COVID19_OPEN_DATA dataset). I hope you enjoy it!