# lecture-notes-lecture-all.pdf - lOMoARcPSD|2980897 Lecture...

• 30

This preview shows page 1 - 4 out of 30 pages.

StuDocu is not sponsored or endorsed by any college or universityLecture notes, lecture allStatistical Data Analysis (Murdoch University)StuDocu is not sponsored or endorsed by any college or universityLecture notes, lecture allStatistical Data Analysis (Murdoch University)Downloaded by Libaan Daud Alin ([email protected])lOMoARcPSD|2980897
MAS183 Stats S2 2011 Chapter 1Sample: data we havePopulaton: data you would likeSamplePopulatonMean (average)(x-bar)μ (mu)Standard Deviatonsσ (sigma)Statstcs is describing data, drawing inferences and knowing the conditons in which the inferences are valid. You can never be certain, but this is controlled by the probability theory. Stats method; ask a queston, get data, analyze data and interpret data. The sample size afects statstcal integrity. Bio. Living, metric = measure. Biometric = blood pressure etc. Normal range-who decides what is normal? Average-where is the upper and lower range? Quanttatve data is not always the most important thing in a conclusion, quality is a factor. Regression to the mean – extreme, values tend to drop back to the mean.Chapter 3Categorical data are labels such as; colour etc. & numerical data is numbers. Numerical coding is less ambiguous to computers. Types of categorical data are as follows;1.Nominal: no order, names2.Ordinal: intrinsic order-disagree, agree, strongly agree (scale)Types of numerical data;1.Contnuous numerical (measurements)2.Discrete (counts) –integers1 row = 1 case as it represents 1 individual’s data. Don’t graph the raw data –use instead tables or spreadsheets. You can graph the distributon of the data. Distributon graphs for categorical data; pie & bar graphs. Bar graph sorted in size order when no intrinsic order. Dot plot to show distributon pictures-numerical contnuous data.An outlier is a piece of data which lies outside of the overall patern data. A should be treated as diferent species, not ignored. Outliers are not always clear cut. Distributon can be clustery, this has implicatons. They can be uni or bimodal. Histograms are always a tradeof between detail and the big picture. The big picture eliminates low or zero frequencies. Small data sets prefer wide columns. Box plots work in quartles- they show how spread the data is. Distributon can be described by;Locaton –the central tendency, where on average is all the data.Downloaded by Libaan Daud Alin ([email protected])lOMoARcPSD|2980897
Spread-the dispersion of the data.Shape-symmetric or skewed, uni or bimodal.Other –extremites*These are in order of reliability.An outlier can be defned by lying outside of 1.5* the interquartle range (1-5 IQR). Measure the locaton: sample mean (average). Sample Standard deviaton (s) measures the spread. To get the standard deviaton:1.Calculate the deviatons from the mean2.Square each of these deviatons and sum up.3.Divide by total number of data idem -1.4.Square root this number.
• • • 