You were young and just discovered that Machine Learning existed. Maybe some friend told you about it because saw it on a movie, or maybe you just clicked on some pop-up that misteriously appeared while web surfing.
You browsed a bit about the topic and inevitably you discover it.
Wow, this Titanic thing looks exciting.
Yes. Titanic dataset. When compared to Iris, it's like Mayweather vs Logan Paul. So you end up installing ScikitLearn locally. You do it nervously, as you're afraid of having some conflicts with your current Python libraries on your dad's computer.
After sweating a bit your T-shirt, you're just in front of something new which was deliberately hidden for most humans. You type model.fit(X,y).
Ecstasy. You've been keeping all this power inside you until this point, it was just a matter of time throwing it out. A new world opens up for you. Now you're a grown up.
You then start wanting more and more. Logistic regression doesn't seem that exciting to you anymore, so you go for Random Forests. You also discover something called XGBoost. This tabular data thing starts to seem boring, so you want to try images. Then video.
You then discover a new category on one of these sites: Neural nets. This changes everything. Nothing will be the same after this point. You forget about ScikitLearn and go for Keras or Pytorch. First, some multi layer perceptron. New dopamine peak reached. But after some epochs, your brain requires something more exciting.
Ten layers aren't deep enough at this point. You need more. You need it deep. You need the biggest neural network architectures. MNIST doesn't excite you at this point and you go to Kaggle looking for huge datasets. Really huge datasets.
You become obsessed about it. You can't stop thinking about Machine Learning. Now your Instagram is full of cool neural network applications. Your twitter feed only talks about new papers and frameworks coming out.
You arrive home and you cannot avoid it. You enter Medium. And you type the following:
Predict bitcoin price with LSTM autoencoder and GANs
At this point you're completely lost. Your brain is addicted to Machine Learning p*rn. And it's like a drug. You're a dopamine junkie.
Some months or years later you start working as a Data Scientist or Machine Learning engineer and you realise that reality is not like p*rn. Dataset sizes are not that impressive. Production models are not that sexy. Maybe one thing is true: you'll be f*cked many times.
I'm starting this project to talk about how Data Science and Machine Learning look like in real life. I'll talk about many issues related to learning and working on the DS/ML field. I've commited many mistakes in the past and I'll make many more.
Of course, you can subscribe in case you're interested in reading more about the truths of being a data scientist in real life. Also, if you share this post on social media I would be extremely thankful.