PHILOSOPHICAL FOUNDATIONS
1. The Nature Scientific Inference
2. Statistical Models
-The mathematical beauty of nature
-The messy part: what is a Probability?
3. Integrating Models and Data
SAMPLING
1. The Nature of Induction
2. Statistical Inference
- Describing the Data
- Estimating the Truth
3. Error and Bias
2. what is
STATISTICS ?Statistics is the science of learning from data, and of
measuring, controlling, and communicating uncertainty
” {Marie Davidian}
Applied philosophy of science
” {OSCAR KEMPTHORNE}
4. PHILOSOPHICAL FOUNDATIONS
1. The Nature Scientific Inference
2. Statistical Models
2.1. The mathematical beauty of nature
2.2. The messy part: what is a Probability?
3. Integrating Models and Data
7. PRAGMATISMREALISM
TRUTH is REAL and science
aims to discover it
TRUTH is in our minds but
it’s still USEFUL
PREDICTION
THEORIES describe
REALITY
THEORIES describe
OBSERVATIONS
9. Give me the positions and velocities of all the
particles of the universe and I will predict the
future
{ P. SIMON de LAPLACE }
”
CLOCKWORK UNIVERSE
10. DETERMINISTIC MODELS
Discovery of Neptune through
deterministic models…
(… sort of)
ERROR FUNCTION to account for observation errors
13. ● Physical Probability
- Frequency: long-run outcome
- Propensity: property of the system
● Evidential Probability (Bayesian)
- Measure of statement uncertainty
WHAT IS PROBABILITY?
25. ● Model fitting and parameter estimation
- Least squares, Maximum Likelihood, Bayesian
- Parameter uncertainty
INTEGRATING MODELS and DATA
26. ● Model fitting and parameter estimation
● Model comparison
- Best fit
- Parsimony and information
- Predictive power
INTEGRATING MODELS and DATA
27. INTEGRATING MODELS and DATA
● Model fitting and parameter estimation
● Model comparison
● Hypothesis testing
- Statistical significance
- Power
28. ● Model fitting and parameter estimation
● Model comparison
● Hypothesis testing
● Prediction
- Repeatability
- Forecasting
INTEGRATING MODELS and DATA
29. ● Model fitting and parameter estimation
● Model comparison
● Hypothesis testing
● Prediction
INTEGRATING MODELS and DATA
31. TECHNIQUESMODELS
● Linear models, GLMs
● Normal distribution
● Null hypothesis
● Hierarchical models, repeated
measures
● Phylogenetic models
● Population models
● ...your own hypothesis
● Least squares regression
● ANOVA
● Maximum Likelihood
● Hypothesis testing (p-value)
● Model selection (AIC, etc)
● Monte Carlo, Bootstrapping
● Goodness of fit (R2
,etc)
● Bayesian Inference
THE BIOLOGY THE SCIENTIFIC METHOD
49. ERROR and BIAS
● Standard Error: Standard Deviation of the
parameter estimates corresponding to different
samples
● Bias: Average difference between estimates and
true parameters among samples
53. MEASUREMENTSAMPLING
BIAS
How random is sampling
BIAS
Technique tends to over or
underestimate
ERROR
How consistent is the
measurement
ERROR
How variable are different
samples
54. What is a CONFIDENCE INTERVAL?
Find out the
definition for
tomorrow