Statistical Concepts

From NovaOrdis Knowledge Base
Revision as of 16:34, 10 October 2016 by Ovidiu (talk | contribs) (→‎Concepts)
Jump to navigation Jump to search

Internal

Concepts

  • mean
  • standard deviation
  • linear regression
  • correlation
  • Bayes Rule
  • Scatter Plot
  • Linearity (linear exact or not exact)
  • Positive and negative linear relationship.
  • Outlier
  • Deviation
  • Noise - deviation from a linear graph.
  • Monotonicity.
  • Bar Charts. Applies to 2D data.
  • Global trends.
  • Historgram. A bar chart where the vertical axis is a frequency count, as a function of the range. Applies to 1D data.
  • Frequency Count
  • Pie charts - represent relative outcomes.
  • Unrelated data
  • Simpson's paradox
  • Be skeptical and really understand how to turn raw data into conclusions.
  • Probability - the opposite of statistics.
  • P() notation
  • Truth table
  • Probability of a composite event (independence)
  • Dependence
  • Conditional probability
  • Conditional probability notation - important for Bayes Rule
  • Total probability
  • Bayes Rule
    • [1]
    • Prior probability
    • Unreliable measurement (Sensitivity/Specificity)
    • Joint probabilty
    • Posterior probabilty
  • Probability Distribution
  • Continous Probability Distribution.
  • Density of probability
  • Correlation vs. Causation
  • Variables
  • Definition of correlation (Is correlation injectivity?)
  • Confounding variable.
  • Estimators
  • Laplacian estimator
  • Empirical (observational) frequency
  • Maximum likelihood estimator

TODO

  • Relocate Continous Functions