WHY WE USE EXPLORATORY DATA ANALYSIS DATA ARE
- Slides: 21
WHY WE USE EXPLORATORY DATA ANALYSIS DATA ARE SAMPLE DATA FROM „NORMAL“ POPULATION? YES ESTIMATES BASED ON NORMAL DISTRIB. NO WHY ? OUTLIERS EXTREMS KURTOSIS, SKEWNESS QUANTILE (ROBUST) ESTIMATES 1 CAN WE REMOVED THEM ? YES NO TRANSFORMATIONS QUANTILE (ROBUST) ESTIMATES TRANSFORMATIONS
METHODS OF EDA Graphical: dot plot box plot notched box plot QQ plot histogram density plots 2 Tests: tests of normality minimal sample size
DOT PLOT 3
BOX PLOT outer fence inner outer inner median číselná osa lower quartil 4 upper kvartil interquartile range (H)
NOTCHED BOX PLOT RF confidence interval of median 5
Q-Q PLOT measured values Y: sample quantiles (ordered values) 6 ideal match between sample values and theoretical distribution Line a=0, b = 1 X: theoretical quantiles (ordered values)of analysed distribution
Q-Q GRAF 7
Q-Q GRAF 8
Q-Q plot left-leaning – skewed to right-leaning – skewed to left platycurtic („flat, broad“) 9 leptocurtic(„steep, slender“)
10
11
HISTOGRAM 12
HISTOGRAM correct width of interval: 13
HISTOGRAM – kernel density function 14
TRANSFORMATION Aim of transformation: reduction of variance better level of symmetry(normality) of data Transformation function: non-linear function monotonic function 15
TRANSFORMATION – basic concept 0. 8 transformed mean and its projection to original data set Transformed data 0. 6 0. 4 0. 2 mean of original data 0 -0. 2 -0. 4 16 0 0. 5 1 1. 5 2 2. 5 Original data (tree-rings widths in mm) 3 3. 5
TRANSFORMATION – logaritmic transformation 17
TRANSFORMATION – power transformation 18
TRANSFORMATION – Box-Cox 19
TRANSFORMATION – Box-Cox 20
TRANSFORMATION– estimate of optimal = 1 is not included in interval estimate of . It means that interval estimate of parameter transformation will be probably successful logarithm of likelihood function for various values of max. LF – 0, 5*quantile 2 optimal 1. 00 21
- Mikael ferm
- Why why why why
- Exploratory data analysis lecture notes
- Eda.a.
- Exploratory factor analysis
- Clustering in business intelligence
- Criteria of evaluating secondary data
- Exploratory research secondary data
- Dont ask why why why
- Why-why analysis
- Contoh laporan root cause analysis
- Types of sae
- Exploratory research design meaning
- Exploratory style vs modern software development
- Exploratory research example
- Examples of conclusive research
- Objectives of literature review
- Where are exploratory robots used
- Exploratory and non directional verb
- Exploratory essay conclusion
- Exploratory vs descriptive vs causal research
- Broad problem area example