Machine Learning

Libraries#

Since Python is dominant in this space, these are all Python libraries unless otherwise specified.

ML#

• tensorflow
• pytorch
• scikitlearn

Math#

• numpy
• Pandas

Visualization#

• matplotlib
• [plotly.py]https://github.com/plotly/plotly.py()
• plotly.js (JS)

Tools#

• Marimo
• List of neural network visualizers

Types#

There are four types of machine learning algorithms:

• supervised
• semi-supervised
• unsupervised
• reinforcement

Algorithms#

K-means#

This is a centroid-based algorithm. As a result it’s simple and fast, but also not great at outliers. Best if you have well clustered/separated datasets, and you’re trying to choose a “group” or “centroid” that a new datapoint is closest to. It just uses the distance to the nearest centroid. So for overlapping clusters we probably don’t want to use this.

https://www.geeksforgeeks.org/machine-learning/k-means-clustering-introduction/ https://www.analyticsvidhya.com/blog/2021/01/a-simple-guide-to-centroid-based-clustering-with-python-code/

Linear regression Logistic regression Decision tree Support Vector Machines Naive Bayes KNN Random Forest Dimensionality reduction algorithms AdaBoost

Neural networks Show less

https://www.reddit.com/r/MachineLearning/comments/16j8at7/d_is_classic_ml_still_relevant/ https://www.ncbi.nlm.nih.gov/books/NBK597496/