> For the complete documentation index, see [llms.txt](https://www.mlcompendium.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://www.mlcompendium.com/predictive-ml/intro.md). # Introduction Each classical learner is one inductive bias, and each bias uses the one before it. After this chapter the reader can pick a linear model, a tree, a cluster, a forecast, or a graph method and say what it assumes. The chapter moves from supervised learners with a label, to unsupervised methods without one, to signals that unfold in time, and ends with graphs. ## Supervised The supervised part starts before there is a label at all. [Data Mining](/predictive-ml/data-mining.md) is about association rules, Apriori, FP-Growth, and how to read support, confidence, and lift when the goal is frequent itemsets rather than a prediction. The first real boundary between classes comes next: [Linear Separator Algorithms](/predictive-ml/linear-separator-algorithms.md) covers support vector machines and related linear separators, with kernels, SMO, libsvm vs liblinear, and overfitting advice. The opposite bias, trusting whatever is close, is [Nearest Neighbors](/predictive-ml/nearest-neighbors.md), the scalable nearest-neighbour libraries and notes on choosing k. Once a model can fit, it can also fit noise. [Regularization](/predictive-ml/regularization.md) is the penalty added to a loss so that such a model is discouraged, with norms, L1/L2 sparsity intuition, and priors. [Probabilistic Models](/predictive-ml/probabilistic-models.md) replaces geometry with probability: naive Bayes, Bayesian networks, Markov models, HMMs, IOHMM, and CRF. When the target is continuous rather than a class, [Regression](/predictive-ml/regression.md) covers the algorithms, kernel regression, and the error metrics used to judge them. Trees give a different bias again. [Decision Trees](/predictive-ml/decision-trees.md) covers Hellinger splits, CART, KD-trees, random forests, and extremely randomized trees, and [Ensembles](/predictive-ml/ensembles.md) then combines many models through voting, bagging, boosting, and stacking, up to gradient boosting, XGBoost, and CatBoost. The last two pages evolve the learner itself: [Genetic Algorithms & Genetic Programming](/predictive-ml/genetic-algorithms-and-genetic-programming.md) contrasts evolving parameters with evolving programs, and [Learning Classifier Systems](/predictive-ml/learning-classifier-systems.md) defines LCS and XCS, where the learner is a rule population rather than a single model. ## Unsupervised Without a label, the question becomes what structure is already in the data. [Clustering Algorithms](/predictive-ml/clustering-algorithms.md) groups unlabeled points, from k-means and GMM through density-based and constrained clustering. [Dimensionality Reduction Methods](/predictive-ml/dimensionality-reduction-methods.md) projects high-dimensional data with PCA, SVD, kernel PCA, LDA, ICA, LSA, and manifold methods such as t-SNE. [Anomaly Detection](/predictive-ml/anomaly-detection.md) separates novelty from outlier detection and runs from elliptic envelopes and LOF through isolation forests and one-class SVM. Logs are a common unlabeled source of their own. [Process Mining](/predictive-ml/process-mining.md) recovers the real process from event logs when the documented one is incomplete or ignored, and [Log Parsing / Templatization](/predictive-ml/templatization.md) turns raw log lines into templates and uses them for log-based anomaly detection. ## Signals Some data has an order in time, and that order is the signal. [Forecasting](/predictive-ml/forecasting.md) covers time-series components, stationarity, decomposition, forecasting methods, and accuracy. [Time Series Search](/predictive-ml/time-series-search.md) compares series with DTW, then clusters them and detects anomalies in them. [Fourier Transform](/predictive-ml/fourier-transform.md) gives the frequency-domain view of a signal, plus wavelets, before modeling, and [Digital Signal Processing (DSP)](/predictive-ml/digital-signal-processing-dsp.md) is the toolbox shelf of SciPy, Librosa, and beat-detection helpers. Audio is the signal case with its own path. [Audio Basics](/predictive-ml/audio-basics.md) is Ketan Doshi's series from sound and spectrograms to speech recognition, [Audio Terminology](/predictive-ml/audio-terminology.md) defines source separation and sound event detection so the task names match the methods, [Audio Feature Engineering](/predictive-ml/audio-feature-engineering.md) covers Mel spectrograms, MFCC, and related features, and [Audio Algorithms](/predictive-ml/audio-algorithms.md) collects the concrete methods and packages for those tasks. ## Graphs The last bias is that the relations between points matter more than the points. [Graph Theory](/predictive-ml/graph-theory.md) covers centrality, community detection, courses, and tools, and [Social Network Analysis](/predictive-ml/social-network-analysis.md) is its applied social-graph companion, with definitions, people, and traits that spread in networks. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://www.mlcompendium.com/predictive-ml/intro.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.