New Post

Chat GPT is pretty legit

OpenAIs LLM is a playful chatbot to use and produces practical advice and code

I recently asked ChatGPT this as I am learning, ramping up, training, and getting certified in various python & machine learning courses. Kaggle, learning python the hard way, Codecademy and Coursera are great resources to get you started. As I’ve learned in my previous roles and time at university, theory only gets one so far. One needs daily practice to ascend to mastery and fine-tune one’s capabilities.

And in the data science, coding, model training, mathematics intensive ML world this is all intertwined, and one needs to be immersed in one’s so-called sandbox. One needs to understand how the algorithms relate to the business need:

"Something I only learned as an AI practitioner was that you have to think about how the model will work in the real-world situation you’re building it for; that context can change the entire end result that you’re shooting for. For example, one of our manufacturing clients may define an item with one dent as still ok, while an item with five dents is defective. We have to figure out how to embed these rules in the network while maintaining a low number of false negatives, and we have to keep the false positives down so the model still produces a cost benefit for the client." ~ Dillon Laird

One needs to also understand practicing and executing algorithms that are ethical.

I wanted to see some basic concrete examples of python code for ML model training:

I quickly inferred that one can leverage the scikit-learn library. I’ve already been exposed and using the Numpy & Panda library. I’ve also learned the specific regression algorithm leveraged here. I’ve seen a lot of linear regression in action, while this is the first time I’m noticing a logistic regression.

I can try out this code in my own IDE (currently running spyder). Although at first glance one may not know how accurate the code is but overall it’s a great place to get started!

Further Reading: