Final

• Random


  Each entity of the population has an equal chance of getting chosen
• Systematic random


  Starting points with fixed intervalues
• Stratified


  Population is divided into groups or strata, sampling occurs within
• Kruskal-Wallis test


  Compares ordinal and non-normal variables for more than two groups
• Outcome variables


  Dependent = plausible results, if plausible then independent not important
• Machine Learning


  Subdiscipline of AI, study of computer algorithms that improve through use and data
• Supervised learning


  Builds mathematical model for data with independent and dependent variables
• Unsupervised learning


  Finds structures of data that contains only inputs
• Decision trees


  Non-parametric supervised learning method used for classification and regression
• Classification trees


  Decision trees where target can have a discrete set of variables
• Regression trees


  Decision trees where target variables can take continuous values
• Ensemble models


  Machine learning algorithms that use an ensemble of decision trees
• Partial dependence plot


  Plot of relationship between predicted values and explanatory variables
• Factor analysis


  Collapses columns of data set to create a smaller number to indicate new linear combinations
• Dimensionality reduction


  Process of reducing the number of attributes in data while keeping substantial amounts of original data
• Ordination


  Any operation of data matrix that reduces the dimensionality, species composition, abundance in each column
• Cluster analysis


  Collapses data row-wise by data that are similar to one another
• Partitioning


  Number of groups and algorithm divided by sample into given number of groups
• Hierarchical


  Determining the best number of groups itself, distinct based on distances
• Eigen analysis


  Tries to find non-correlated linear combinations of original variables, new variables to explain variances within data by axis
• Direct gradient


  Uses environmental variables to determine environmental gradients
• Logistic regression

  Used when the dependent variable is binary
• Log Pseudo-likelihood

  Measures how well the model fits the data, only meaningful/comparable within each species
• Model Likelihood Ratio Chi-Square test + P-value

  Tests how other models fit and the Probability of Obtaining the Chi-squared test
• Pseudo R-Squared

  Logistic Regression does not have an R-squared as in OLS regression
• Odds ratio regression coefficient

  Expressed as Probability (ratio of Probability of Presence to Probability of Absence)
• Non-Parametric tests

  Distribution or free tests, do not assume anything about the underlying distribution
• Count data

  Discrete and bound by Zero, no negatives so you cannot get over dispersion
• Over dispersion

  Data with variables higher than expected
• Generalized Linear Models

  Umbrella for wide range of regressions