I'm an undergrad presently in my senior year 1st semester and my college timings are a bit different from the other colleges its from 0900hrs to 1730hrs and that's some hectic schedule where our college is a tire 3 college and we are just going to college to strictly waste our time. can you just imagine, the 1/3rd part of your day getting wasted just like that, that's why my major progress will be on weekends, as studying MLops is not the only thing I practice, I'm building new habits of reading books, meditating and an hour goes into journaling every day. So yeah, today I checked the syllabus of machine learning and gave a deeper look into unit 1 and as I have some prior knowledge in machine learning I came through all the basic concepts in machine learning like how it evolved. the categories? the applications? etc.
I'll be sharing all the resources i've used to learn this topic from, in the blog itself . so that even you can refer them and learn . That way you wont need to worry about the best article to learn. 🕊️
ML is categorized broadly into 3 types supervised, unsupervised and reinforcement learning. Supervised is where you give the data and the target labels associated with the data, to the model, and the model gets trained on it and performs classification, and regression. here everything is given by the developer itself and every aspect of the training is supervised by the developer. Unsupervised is where the developer gives just the data and the model classifies the cluster by itself by studying the features of the cluster. and reinforcement learning is where the model is taught on the basis of reward and penalties where the model is give a reward score when it reaches closer to the goal state and gets penalties when it deviates from the route. these were broadly the categories of Machine Learning.
✨ Reference link: towardsdatascience.com/types-of-machine-lea..
Now coming to the next topic Evaluation metrics. Well, I was thinking for a long time to gain some legit knowledge on this topic as most ML developers just forget about this topic and directly give the model the unknown data to test the model, it's always important to evaluate the model first. now there are many different methods. some of them are
Accuracy
Precision
Recall
F1-score
Note:
the precision and recall trade must be understood and some other similar topics like bias and variance tradeoff, these are various topics to get knowledge on before getting further understanding about these metrics
there are other metrics like ROC curve, confusion matrix, etc. but the basic lies in these.
✨ Reference link: https://www.analyticsvidhya.com/blog/2019/08/11-important-model-evaluation-error-metrics/
✨ Reference link: https://www.analyticsvidhya.com/blog/2020/09/precision-recall-machine-learning/#Precision_and_Recall_Trade-off
✨ Reference link: https://www.analyticsvidhya.com/blog/2020/08/bias-and-variance-tradeoff-machine-learning/?utm_source=blog&utm_medium=precision_and_recall
Just refer to the above links to gain better knowledge on the topic and let's move ahead in the topic. I'll add my notes down below. that's all for today, now I have a better understanding of the different evaluation matrices and how they work. if you are with me in this journey follow me on Twitter @lokstwt and subscribe to this newsletter to get later updates as I move ahead in this journey. Share it with more of your friends who are looking forward to getting into MLops.
Thank you ✌🏾
Today Note's 📝:
