rewrite-toc

TODO (Learn)
    * Frequency Analysis - PSD, Band power, Frequency domain
    * Singular spectrum analysis -> Taken's embedding, applies PCA to time lags 
in phase space to separate signal into stationary components
    * Correlation Dimension - Implement. Very beautiful concept
    * Recurrence plot - Implement. Seems very reasonable approach to projecting 
multidimensional data. 
    * heteroskedasticity models

TODO (Find Home)
    * ANOVA
    * Parameter jiggling to get smooth estimators (read pub)

Probability Theory
    - Moments
    - Empirical Distributions, CDF
Resampling
    - IID methods: Bootstrapping, Permutation
    - NON IID: (AKA Surrogate data)
Monte Carlo Methods
    - Expectation Values
    - Integrals
Hypothesis Testing
    - Parametric Tests
        - T, F, Chi2, Binomial, etc
    - Non-Parametric Tests
        - Ranksum/Signrank
    - Permutation-testing
        
    - Composing p-values
Inference: Non-parametric:
    - Static Relation metrics: Corr, Spr, MI
    - Dynamic Relation metrics:
        - Complexity (PI)
        - Stationary (CrossCorr, GC)
        - Semi-Stationary (TE)
    - Information Theory
        - Bulk Metrics: H, TC
        - Higher order relations: MMI, PID, AIS
    - Graph-Theoretic metrics: Degree, CC, etc.
    - Time series analysis:
        - Stationarity testing
        - Frequency analysis
        - Phase-space methods
            - Dim. Red. in space and time simultaneously
            - Taken's embedding
Inference: Parametric
    - Classification: Logit, Cross-Entropy
    - Regression: Linear, Logistic, GLM
    - Time series analysis
        - AR/VAR/MAR, MA, ARMA
        - Forecasting
    - Model selection
        - ROC/AUC
        - Cross-validation
        - Model Selection Criteria: AIC, BIC, BMS(Bayes factors)
    - Regularization
        - L1, L2
    - Generative Models
Dimensionality Reduction:
    - PCA, FA, NMF, tSNE
    - Autoencoders