In this post, I briefly go over the concept of an unsupervised learning method, mean shift clustering, and its implementation in Python — Mean shift is an unsupervised learning algorithm that is mostly used for clustering. It is widely used in real-world data analysis (e.g., image segmentation)because it’s non-parametric and doesn’t require any predefined shape of the clusters in the feature space.

In this post, I briefly talk about how to understand an unsupervised learning method, OPTICS, and its implementation in Python. — OPTICS stands for Ordering points to identify the clustering structure. It is a density-based unsupervised learning algorithm, which was developed by the same research group that developed DBSCAN. As discussed in my last post, DBSCAN has a major disadvantage in that it struggles to identify clusters in data of varying…

In this post, I briefly go over the ideas of DBSCAN and its implementation in Python. — DBSCAN stands for Density-Based Spatial Clustering of Applications with Noise, which is an unsupervised learning algorithm. DBSCAN is one of the most widely used clustering methods because the clusters found by DBSCAN can be any shape, which can deal with some special cases that other methods cannot. One of the…

In this post, I briefly go over the idea of a dimension reduction tool, random projection, and its implementation in Python. — Dimension reduction is usually a must-to-do preprocessing when dealing with big data. One of the most widely used methods is Principal Component Analysis (PCA), but the major shortcoming of PCA is that it could be very computationally expensive for high-dimension data. …

In this post, I briefly go over the concept of an unsupervised learning method, Fuzzy C-Means clustering, and its implementation in Python. — Fuzzy C-means clustering algorithm is an unsupervised learning method. Before learning the details, let me first decipher its fancy name. So, “fuzzy” here means “not sure”, which indicates that it’s a soft clustering method. …

In this post, I briefly go over the concept of an unsupervised learning method, the Gaussian Mixture Model, and its implementation in Python. — The Gaussian mixture model (GMM) is well-known as an unsupervised learning algorithm for clustering. Here, “Gaussian” means the Gaussian distribution, described by mean and variance; mixture means the mixture of more than one Gaussian distribution. The idea is simple. Suppose we know a collection of data points are from a…

This post discusses basic knowledge about Smoothing Splines and implementations in R. — One of the most frequently asked questions about regression splines is “How to determine the number and positions of knots?” A lot of methods out there can address this topic, but there’s a shortcut to avoid considering such questions in the modeling. It’s the smoothing splines. Concept of Smoothing Splines Instead of requesting a…

This post discusses basic knowledge about Regression Splines and implementations in both R and Python. — The linear model is named so because of the linear relationship between the input (independent variable) and the output (dependent variable). Even though we know it’s of a high probability that the real-world data shows nonlinearity, people usually keep regarding the linear model as one of the top choices.

An end-to-end analysis example in R about using random effects in linear models. — The simple linear model has an important assumption, the independence of the observations. This assumption holds in most carefully designed experiments but seldom does in real-life datasets. One of the biggest risks of assuming correlated data as independent is that your linear model will always give you a beautiful p-value…