Why Did This Library Was Created In The First Place

Section 1: Datapredict - The Backup Plan When Everything Falls To Shit

With the writing on the wall regarding to my future career prospects on my final year of the second semester, I needed anything to fix my trajectory.

Anything.

Luckily enough, during my second year of university, I had a group software project that relates to Roblox Game Development. Technically, I took up the leadership position after our previous leader tend to underestimate and procrastinate on this group project to the point that we left with one week for prototyping despite me pointing out repatedly. That girl cried when the deadline is near and I ended up irritated, before taking up the leadership position given that I had very casual Roblox game development experience. Somehow, we ended up completing the game development in a year. I then wanted to commercialize this game, so I ended up working alone on this on the final year given that my grades were failing anyways.

Now… As I scroll through the Roblox DevForum, I noticed that there was a gap in machine learning resources. At best, there was a guy called Kironte who created a library for simple neural network. It’s nice for beginners who want to just run neural networks on Roblox, but it isn’t quite optimized for live Roblox games. Not to mention, where are the other machine learning models?

This has become a basis for my prototype project during my final year of the second semester: Machine And Deep Learning Library (MDLL). It had like linear regression, logistic regression and K-Means models only. But, I published the post for this and look for demand.

Apparently, there was. A small amount of people struggling to use it and wanting to apply it to their games. So, I ended up developing the beginning of a true machine learning library written in pure Lua.

I had multiple names for my library as my previous library’s name was too raw and crude. I looked at them online and made sure nobody already took it, regardless if it was a machine learning product or not. I ended up settling with DataPredict™, which isn’t quite a bad name actually.

Now, this is where my software engineering knowledge from my university’s first year software engineering classes becomes useful: inheritance, composition, design patterns, SOLID and DRY. This forms the basis of my DataPredict™ library and a first release made to the public through Roblox DevForum. At the same time, I also asked the moderators to delete the MDLL’s Roblox DevForum post so that they don’t go to the outdated resources. Frankly, I should have actually link it to DataPredict™’s Roblox DevForum post instead.

It started off with a normal machine learning library in the first year. People don’t really understand the use cases. Me neither. All I know is that I want to be the first one to have a comprehensive machine learning library for Roblox. Someone criticized the poor documentation back then and I realized he was correct as I was too excited in releasing the DataPredict™ library. So I ended up doing a better job in documentation, which ended up helping DataPredict™’s reputation and ended up paying off in the long term.

Then the introduction of deep reinforcement learning library in the second year. This came from people asking for self-learning AIs. I ended up accepting it and built it.

Then the introduction of statistical models and tabular reinforcement learning that got mixed inside DataPredict™. This is where I literally understood the use cases for machine learning models and ended up building a huge list of tutorials. I will literally write a tutorial by the time an idea pops up into my head, provided that it was a genuine use case.

During my first year of DataPredict™’s development, it ended up boosting my chances of getting non-existent Roblox game developer job in Malaysia. Although it wasn’t quite high, at least I know my trajectory isn’t that quite fucked.

Except… I severely underestimated the impact of my untreated mental could impact my work performance… I may or may not have broken several codebases as a result of stiffness and haziness from my brain while working, as well as hallucinating some codes and some instructions.

At the very least, let me try one last time in succeeding.

I mean… DataPredict™ having over 136K views and 268 likes for over three years is something. Being top 40 in terms of most viewed community resource is also something. Over hundreds of replies is also something. My direct competitors not being able to outcompete me and falls into obscurity due to DataPredict™ being so visible is also something. People demanding for DataPredict™ to be free for them while DataPredict™ explicitly have commercial licenses is also something. Having a monopoly position for a machine learning library written in pure Lua is also something.

At the very least, I now know that I can build something valuable in the midst of my parents’ stupidity and hypocrisies, as well as lacking owning up to their mistakes and hypocrisies.

That being said… DataPredict™ is almost being close to done after all these years. I might eventually put this into a perpetual maintenance mode until people demand me to add new stuff. What a great journey, DataPredict™.

Once DataPredict™ is done, I’m going back to:

Drawing lewd anime girl characters and improve it. I had like 2-3 years worth of casual drawing experience here. Mostly sketches, but I did get into digital drawing with a digital tablet. I just want my drawing skills back after years of hell I fucking endured by my father’s LIES. I was this fucking close in being able to draw sketches perfectly on paper. I even managed to get the basics down for my digital drawing using my drawing tablet. I can’t believe that I lost my ability to draw because of my father’s LIES. I just want my drawing skills back after dealing with my father’s LIES. I just want to go back to drawing digitally on the Krita software after dealing with my father’s LIES.
Learn piano again as a beginner.
Binge watch anime since I never had any throughout A-levels to the development of DataPredict™.
Go play games and try out the games’ skin mods for the first time. This is probably the reason why the recent releases of DataPredict™’s updates are so slow. I literally need to collect characters I like in terms personality and appearance so that I can use the games’ skin mods on them.

Also, I just realized the irony of all this…

Despite DataPredict™ being so heavy in mathematics and computer science, I somehow managed to be able to develop DataPredict™ despite my low grades in mathematics and computer science overall. I guess that my A-levels mathematics teacher’s hug and that after-school classes of his signalled to me: I am willing to put effort in you even if I know you’re crying from thinking that you’re about to fail.

I just… didn’t realized that I had internalized his signal to something like this to myself: Keep going. Even if you feel like everything is going to shit, just keep going.

This is probably the reason why I ended up being comfortable receiving criticisms. This also probably made me comfortable with implementing mathematics that I never learnt in my A-levels or the university at all to DataPredict™. This also probably made me comfortable to fail, knowing it is just a setback in life like everything in life.

I still hate my parents though. I will never forgive them at all. Not even in their deathbeds.