MATH1015 Biostatistics 2.1 Week 2/3 MATH1015 Biostatistics Boxplots (P.15-16) Week 2/3 Example: Consider the following data set of 13 observations xi from the previous example: Recall that observations outside the interval (LT,UT) are called outliers or abnormal observations, where 4 6 6 7 7 9 10 11 13 15 22 24 30 Lower threshold value (LT) = lower quartile - 1.5 × IQR 1. Find LT and UT for this sample. Upper threshold value (UT) = upper quartile + 1.5 × IQR. 2. Identify any outliers if they exist. A popular (box type) graphical representation of the following information from a data set is known as a boxplot: 3. Draw a boxplot for this sample following the steps: • Quartiles Q1 , Q2 and Q3 , (draw a rectangular box from the quartiles Q1 to Q3 and mark Q2 within this box) • Smallest and largest observations within (LT,UT), (a) Draw a rectangle (horizontal or vertical) of arbitrary width from Q1 to Q3 . (b) Draw a dotted-line across the rectangle at Q2 . (c) Draw two lines (called, Whiskers) to and from the observations within (LT,UT) from the above rectangle. • Outliers, if exist. Diagram: Suppose that a data set contains three values below the LT (left outliers) and two values above the UT (right outliers). Now we show these information in the diagram below: (d) Mark any identified outliers by ◦ Solution: 1. From the previous example, we have calculated: Median, Q2 = 10; Lower quartile, Q1 = 7. Upper quartile, Q3 = 15. Hence IQR = 8; LT = -5; UT = 27. Boxplots show the shape of the distribution of data very clearly and are helpful in representing any outlying (or extreme) values of a data set. SydU MATH1015 (2015) First semester 1 2. All observations in the interval (-5,27) are considered “legitimate”. Clearly, there is only one data point outside this interval. Therefore, the last observation 30 is considered as abnormally high. This is an outlier. SydU MATH1015 (2015) First semester 2 MATH1015 Biostatistics Week 2/3 MATH1015 Biostatistics 3. The following boxplot summarises the above information as a graph indicating the outlier by o : Week 2/3 Notes: • Boxplots are useful to compare a continuous variable (e.g. length, weight etc) with a nominal variable (e.g. treatment). • Length of whisker in R is by default chosen to be 1.5×IQR, • Boxplots give a simple visual display and hence a quick impression of the shape of the data set: – Symmetrical: left and right tails are similar Boxplot in R: R can be used to draw a boxplot. Let x contains the data. – left skewed: boxplot is stretched to the left. > x=c(4,6,6,7,7,9,10,11,13,15,22,24,30) > boxplot(x) – Right skewed: boxplot is stretched to the right. Now we look at a number of additional summaries from a data set. LT min Q1 Q2 2.2 2nd max UT max Q3 Measures of Location and Spread (P.9-11) Measures of Location -5 4 7 5 10 15 10 15 24 20 SydU MATH1015 (2015) First semester 27 25 We have seen that median is a measure of the center of a data set. Another popular measure of the center of a data set is known as the mean. Recall from your high school work that the mean of (4,7, 9, 5, 3) is 4+7+9+5+3 = 5.6. Use your calculator to check 5 this answer. Now we develop this concept to handle common problems instatistics. we use the following notation: 30 30 3 SydU MATH1015 (2015) First semester 4 MATH1015 Biostatistics Week 2/3 MATH1015 Biostatistics A Notation Solution: Suppose that we have n observations from an experiment. This collection (or set) of n values is called a sample. Let x1 be the first sample point or observation; x2 be the second sample point or observation etc and xn be the nth sample point or observation. 4 X Example: Suppose that we have a sample of five observations {4, 7, 9, 5, 3}. For this sample, the first observed values is 4 and therefore we write x1 = 4 to identify it. Similarly, x2 = 7, x3 = 9, x4 = 5, x5 = 3. Summation Notation: For simplicity, the sum of these n values x1 , x2 , · · · , xn is abbreviated by the sigma notation as follows: n X xi = x1 + x2 + · · · + xn . i=1 3 X Week 2/3 xi = x1 + x2 + x3 + x4 = 3 + 4 + 5 + 1 = 13 xi = x2 + x3 = 4 + 5 = 9 i=2 4 X (2xi + 3) = (2x1 + 3) + (2x2 + 3) + (2x3 + 3) + (2x4 + 3) i=1 = (6 + 3) + (8 + 3) + (10 + 3) + (2 + 3) = 9 + 11 + 13 + 5 = 38. 4 X x2i = x21 + x22 + x23 + x24 = 32 + 42 + 52 + 12 = 51 i=1 2.2.1 The Sample Mean, p9 i=1 Note: Many calculators use this notation. Please check your calculator now. Example: Consider the Psample: x1 = 4, x2 = 7, x3 = 9, x4 = 5, x5 = 3. Write down 5i=1 xi and evaluate it. Solution: 5 X xi = x1 + x2 + x3 + x4 + x5 = 4 + 7 + 9 + 5 + 3 = 28 The sample mean is the simple arithmetic mean or the average of observations. For n observations x1 , x2 , . . . , xn , this is denoted by x̄ (called x bar) and is given by n x1 + x2 + . . . + xn 1X x̄ = xi . = n n i=1 Example: i=1 Example: Evaluate the following summation expressions for the values (3, 4, 5, 1): 4 4 3 4 X X X X x2i . (2xi + 3) and xi , xi , The mean of the sample of 4 values from a previous example is SydU MATH1015 (2015) First semester SydU MATH1015 (2015) First semester i=1 i=2 i=1 x̄ = 3+4+5+1 = 3.25 . 4 i=1 5 6 MATH1015 Biostatistics Week 2/3 MATH1015 Biostatistics Exercise: Look at your calculator now. Change the mode of your calculator to ’stat’ or ’sd’ or as per calculator instructions. Chack the above answer using your calculator. Note: The mean is very sensitive to large or small outliers in the sample. In such cases it is better to use the median as a measure of the “centre” of the data. Sample Variance and Standard Deviation, p12 In order to motivate this topic, consider the following two sets of observations: 2, 5, 15, 20, 38 12, 13, 15, 19, 21 s s s ss s s s s s x̄ = 16 Use of R R can be used to find the mean of a sample. Practice this example. > x=c(3,4,5,1) > mean(x) >3.25 Exercise: Find the median, mean and mode for the data set: 13.3, 10.7, 11.0, 11.1, 12.9, 11.8, 11.9, 12.2, 10.8, 12.2, 11.6, 11.8 Solution: Order the data xi to find the median: 10.7, 10.8, 11.0, 11.1, 11.6, 11.8, 11.8, 11.9, 12.2, 12.2, 12.9, 13.3 Ans: mean= 11.775; median = 11.8; mode=11.8 and 12.2 In this case, the mode is not unique. Such datasets are also called bimodal. Exercise: Check the mean of this sample using your calculator (now) changing the mode to stat. Exercise: Check the answers using R. SydU MATH1015 (2015) First semester 2.2.2 Week 2/3 7 It is easy to verify that both sets have the same centre or the mean at x̄ = 16. However, the two samples visually appear radically different. This difference lies in the greater spread or variability, or dispersion in the first dataset than the second. Therefore, we need a universal measure to find an indication of the amount of variation that a data set exhibits. We will now describe the most popular measure of spread used in practice known as the sample variance based on n observations. The Sample Variance The difference between an observation and the sample mean is known as the ’deviation of the observation’ from the sample mean. For example, in sample 1 the deviations from the mean are: 2 − 16 = −14, 5 − 16 = −11, 15 − 16 = −1, 20 − 16 = 4, 38 − 16 = 22. The sum of squared deviations divided by 4 is considered as a good measure of the spread and known as the sample variance. For the above sample 1: 2 2 2 2 2 the variance= (−14) +(−11) 4+(−1) +4 +22 = 818 = 204.5. 4 Similarly, for the sample 2, the variance is 15. As seen from the data, the sample 1 has more variablity than the sample 2. SydU MATH1015 (2015) First semester 8 MATH1015 Biostatistics Week 2/3 MATH1015 Biostatistics Week 2/3 12 Calculation of the Sample Variance For a set of n observations x1 , x2 , . . . , xn , the sample variance s2 is given by n 1 X 2 (xi − x̄)2 . s = n − 1 i=1 Note: It is easier to use the following calculation formula in practice. It can be shown after expanding the square term (xi − x̄)2 and re-arranging the terms that the above is equivalent to: !2 " n # n n X X X 1 1 1 or x2 − xi = x2 − nx̄2 . s2 = n − 1 i=1 i n i=1 n − 1 i=1 i Note: You do not need to memorize this formula as it is provided on a formula sheet available from the course web site. Note: The above value is in squared units 1X 689 xi = = 57.4167 • Mean: x̄ = n i=1 12 • Variance: P X 1 ( xi )2 (689)2 1 2 s = xi − = 40095 − = 48.629 n−1 n 11 12 2 Standard Deviation of a Sample It is clear that the sample variance has squared units. Therefore, its square root will provide value in original units. This square root is known as the sample standard deviation. Example: Find the standard deviation of the above sample. Solution: Simply take the square root of the variance. Thus, the Standard Deviation is: √ s = 48.6288 = 6.9734 Example: Find the mean and variance of the sample: Notes: 55, 48, 59, 64, 65, 57, 58, 41, 57, 59, 64, 62 • Many scientific calculators and computer packages (including R) can be used to find the standard deviation of a given dataset. Solution: n = 12. First calculate 12 X i=1 12 X i=1 x2i xi = 55 + 48 + 59 + · · · + 62 = 689 2 2 2 • Look at your calculator now: – Change the mode of your calculator to STAT (or similar depending on your calculator). 2 = 55 + 48 + 59 + · · · + 62 = 40095 SydU MATH1015 (2015) First semester 9 SydU MATH1015 (2015) First semester 10 MATH1015 Biostatistics Week 2/3 – Look for buttons x̄, s2 or σ 2 . Many calculators have 2 s2n−1 or σn−1 button for the sample sd. Check with the user manual for details. • It can be proved that after a change in origin of a data set, the variance and standard deviation remain the same. If the sample points change in scale by a factor c, then the variance changes by a factor of c2 and the sd changes by a factor of c. Exercise: Consider the data set 110, 96, 118, 128, 130, 114, 116, 82, 114, 118, 128, 124. Show that the mean, variance and sd respectively are (approx) 114.84, 194.52, 13.95. Note: the second data set is twice the first and hence the second mean is twice the first mean; second variance is four times the first variance and second sd is twice the first sd. 2.2.3 The Coefficient of Variation The coefficient of variation, denoted CV, is the ratio of the standard deviation to the mean. For a dataset with x̄ 6= 0, we define CV = s x̄ MATH1015 Biostatistics Week 2/3 Example: The CV for the previous dataset is CV = 6.973 s = = 0.1214449 x̄ 57.417 or the s.d. accounts for 12% of the mean. Note: It is clear that the CV is dimensionless as it is a proportion. For example, it is not affected by multiplicative changes of scale. Therefore, the CV is a useful measure for comparing the dispersions of two or more variables that are measured on different scales. The next section considers the corresponding results for grouped data. 2.3 Grouped Data (P.16-17) Recall that large datasets can be summarised with a suitable frequency distribution table with k groups or intervals or bins like this: Group/Class interval y1 < x ≤ y2 y2 < x ≤ y3 .. . Class center u1 = (y1 + y2 )/2 u2 = (y2 + y3 )/2 .. . Frequency f1 f2 .. . Relative frequency f1 /n f2 /n .. . yk < x ≤ yk+1 TOTAL uk = (yk + yk+1 )/2 fk n fk /n 1.000 This ratio of the standard deviation to the mean is a useful statistic for comparing the degree of variation from one data series to another, even if the means are drastically different from each other. Now we look the problem of calculating the mean and variance from such a frequenccy table. SydU MATH1015 (2015) First semester SydU MATH1015 (2015) First semester 11 12 MATH1015 Biostatistics 2.3.1 Week 2/3 The mean of Grouped Data MATH1015 Biostatistics Week 2/3 Solution: n = 35 (the number of values) Suppose that we only have the information provided by a grouped frequency table for a data set. That is, we only have access to the published report and not the original data set. Let k be the number of bins (groups or intervals) and u1 , u2 , . . . , uk be the centres of each interval with corresponding frequencies f1 , f2 , . . . , fk . Then an approximate sample mean is given by 6 X fi ui = 2(99) + 5(109) + 11(119) + 10(129) + 3(139) + 4(149) = 4355 i=1 6 4355 1X fi ui = = 124.4286 ⇒ x̄ = n i=1 35 Exercise: Find the exact mean of the data and compare it to the above approximation. k 1X fi ui . x̄ = n i=1 Example: Consider the data on weight in pounds (recorded to the nearest pound) of 35 female students from week 1. Answer: Using the complete data, check with your calculator and R , sum of all 35 values=4333 and hence the exact mean, x̄ = 123.8. Females: 140 120 130 138 121 125 116 145 150 112 125 130 120 130 131 120 118 125 135 125 118 122 115 102 115 150 110 116 108 95 125 133 110 150 108 Note: The grouped mean and the exact mean are close to each other. We have the frequency distribution from last week: For data from a frequency table, the grouped sample variance is: CLASS INTERVAL 94-104 104-114 114-124 124-134 134-144 144-154 TOTAL CLASS CENTER 99 109 119 129 139 149 2.3.2 FREQUENCY 2 5 11 10 3 4 35 k 1 X fj (uj − x̄)2 s = n − 1 j=1 2 or equivalently " k " k # # k X X X 1 1 or 1 fj u2j − ( fj u2j − n(x̄2 ) . fj uj )2 = s2 = n − 1 j=1 n j=1 n − 1 j=1 Find the grouped mean. SydU MATH1015 (2015) First semester The Variance of Grouped Data 13 SydU MATH1015 (2015) First semester 14 MATH1015 Biostatistics Week 2/3 Example: Find the sample variance from the previous frequency distribution table of 35 female students. Solution: 6 X i=1 fi u2i = 2(992 ) + 5(1092 ) + 11(1192 ) + · · · + 4(1492 ) = 547955 1 ⇒s = (547955 − 35 × 124.4292 ) = 178.3776 34 √ ⇒s = 178.3776 = 13.35581 2 Thus s2 = P 34 x)2 /35 = 542505−43332 /35) 34 CLASS CENTER 129 143 157 171 185 199 213 FREQUENCY 6 17 17 7 8 1 1 57 Answer: Grouped mean=157.2456 and Exact mean=9021/57 = 158.2632 and solution: Check with your calculator and R the following: P P 2 x = 4333; x = 542505. x2 −( CLASS INTERVAL 122-136 136-150 150-164 164-178 178-192 192-206 206-220 TOTAL Week 2/3 grouped variance=367.4431. sd=19.16881. Example: Find the exact sample sd and compare with the grouped sd=13.35581. P MATH1015 Biostatistics Exact variance=(1447141−90212 /57)/56 = 347.3045. sd=18.63611. Additional worked example: Consider the two samples: = 178.8118 and sd=13.37205. Notice that these two values are also close to each other. Sample 1, x: 1.76, 1.45, 1.03, 1.53, 2.34, 1.96, 1.79, 1.21 For Sample 2, y: 0.49, 0.85, 1.00, 1.54, 1.01, 0.75, 2.11, 0.92 each of the two samples, 1. calculate the mean and the standard deviation, Exercise: Using the following frequecy table for 57 male students from week1 (p14), compute the grouped mean and sd using your calculator and R. Compare them with exact values. 2. find Q1 , Q2 , Q3 , LT and U T, 3. find CV, 4. draw both boxplots on the same page. SydU MATH1015 (2015) First semester 15 SydU MATH1015 (2015) First semester 16 MATH1015 Biostatistics Week 2/3 Solution: In ascending order: 1. We have n = 8 is even and 8 8 8 8 P P P P xi = 13.07, x2i = 22.5873, yi = 8.67, yi2 = 11.2153 i=1 i=1 i=1 i=1 Sample 1: 8 X Sample 2: Q1 = 0.80; Q2 = 0.96; Q3 = 1.28. IQR = Q3 − Q1 = 1.28 − 0.80 = 0.48; Since the max = 2.11 lies outside (LT,UT) = (0.08,2.00). 3. CVs are 0.258 and 0.472 respectively. 4. min Q1 Q2 0.49 0.80 0.96 Q3 2nd max UT max 1.28 1.54 2.00 2.11 2 LT 1 0.08 0.51 8 8.67 1X = 1.08 yi = The mean ȳ = 8 i=1 8 v u !2 8 8 u X X 1 u 1 The sd sy = t yi2 − yi 8 − 1 i=1 n i=1 s 8.672 1 11.2153 − = 0.51 = 7 8 Sample 1 xi : Sample 2 yi : Sample 1: Q1 = 1.330; Q2 = 1.645; Q3 = 1.875); IQR = Q3 − Q1 = 1.875 − 1.330 = 0.545; LT = Q1 − 1.5 × IQR = 1.330 − 1.5(0.545) = 0.5125 UT = Q3 + 1.5 × IQR = 1.875 + 1.5(0.545) = 2.6925 There is no outlier. UT = Q3 + 1.5 × IQR = 1.28 + 1.5(0.48) = 2.00 13.07 1 xi = = 1.63 8 i=1 8 v u !2 8 8 u X X 1 u 1 = t xi x2i − 8 − 1 i=1 n i=1 s 13.072 1 = 22.5873 − = 0.42 7 8 Sample 2 : 2. Week 2/3 LT = Q1 − 1.5 × IQR = 0.80 − 1.5(0.48) = 0.08 ; The mean x̄ = The sd sx MATH1015 Biostatistics 0.5 LT 1.33 min 1.65 1.88 1.5 Q1 2.34 2.69 2.0 Q2 Q3 max UT R commands: mean(x) sd(x) sort(x) median(x) sd(x)/mean(x) #cv fivenum(x) boxplot(x,y) #2 boxplots side by side 1.03, 1.21, 1.45, 1.53, 1.76, 1.79, 1.96, 2.34 0.49, 0.75, 0.85, 0.92, 1.00, 1.01, 1.54, 2.11 SydU MATH1015 (2015) First semester 1.03 1.0 where x and y are vectors of measurements. 17 SydU MATH1015 (2015) First semester 18 MATH1015 Biostatistics Week 2/3 In order to develop further concepts and applications of biostatistics, it is convenient to understand the basic theory of probability. Now we look at this topic. 3 An Introduction to Probability Theory and Applications, P29 MATH1015 Biostatistics Week 2/3 1. Toss a fair six-sided die once and observe the number that shows on top. 2. Take a marble from a bag containing 2 red, 1 black and 1 white balls and observe its colour. It is clear that in these random experiments, one cannot state (before the experiment) what a particular outcome will be at each throw. However, we can make a list of all possible outcomes. This chapter considers the following topics: For example: • Basic terminology, • Theory of sets and Venn diagrams, 1. In 1, we observe one of: 1 or 2 or 3 or 4 or 5 or 6. • Probability axioms and counting methods, 2. In 2, we observe one colour from: red or black or white. • Conditional probability and independence. Now we provide the following definition for later reference: Preliminaries • The word fair or unbiased is regularly used in many life science situations. This means that all possible outcomes of an experiment have the same chance to occur. • Any experiment to collect information is called a random experiment, if we are not certain or cannot predict of its outcome(s). It is clear that in a random experiment, one cannot state (before the experiment) what a particular outcome will be. Note: On contrary, a deterministic experiment yields known or predictable outcomes when repeated under the same conditions. Definition: The collection (or the set) of all possible outcomes of a random experiment is called the sample space. This is denoted by S or Ω and be written as S = {· · · }. For example, 1. in experiment 1 above, S = {1, 2, 3, 4, 5, 6}. 2. in experiment 2 above, S = {R, B, W }. The following terminology will be useful in many applications: Definition: An event of a random experiment is a collection of outcomes with specified or interested features. For example, consider the following experiments: SydU MATH1015 (2015) First semester 19 SydU MATH1015 (2015) First semester 20 MATH1015 Biostatistics Week 2/3 Example: List the event A of observing a number less than 3 in experiment 1 above. Ans: A = {1, 2}. Example: A card is selected at random from a box containg 10 cards with numbers 1 to 10. List the events: A of observing even numbers and B of observing numbers divisible by 4. Ans: A = {2, 4, 6, 8, 10}; 3.1 B = {4, 8}. Probability of equally likely outcomes/events First consider the concept of equally likely outcomes. Equally Likely Outcomes: The outcomes of a random experiment (or in a sample space) are called equally likely if all of them have the same chance of occurrence. In a historical note, the probability was considered as the chance of an event to occur which expresses the strength of one’s belief. Therefore, this was known as subjective probability. However, this was later developed with a number of common concepts including equally likely outcomes. Therefore, we have the following definion: Definition: The probability of an event A is the relative frequency of its set of outcomes over an indefinitely large number of repeated trials under identical conditions. This is denoted by P (A). MATH1015 Biostatistics Week 2/3 Calculating Probabilities Suppose we have a random experiment, which has exactly n total possible equally likely outcomes. Let A be an event of interest within this sample space containing m number of simple outcomes. Then the probability assigned to A, P (A) is given by: m P (A) = . n Examples: 1. Throw a fair six-sided die. There are 6 equally likely possible outcomes. The sample space, S of this experiment is S = {1, 2, 3, 4, 5, 6} . If A denotes the event of observing an even number, then A = {2, 4, 6}. Prob(an even number) = P (A) = 3 1 = . 6 2 2. Toss a fair coin 3 times. There are 8 possible equally likely outcome and the sample space is S = { HHH, HHT, HTH, THH, TTH, HTT, THT, TTT } . • Let A be the event of observing exactly two heads in this experiment. Then A = { HHT, HTH, THH } and the probability of observing exactly two heads is P (A) = ..... SydU MATH1015 (2015) First semester 21 SydU MATH1015 (2015) First semester 22 MATH1015 Biostatistics Week 2/3 • Let B be the event of observing at least one head. Then the event is B = { HHH, HHT, HTH, THH, TTH, HTT, THT }. Hence, the probability of observing at least one head is P (B) = ..... 3.2 Probability using tree diagrams, p33 MATH1015 Biostatistics Example: A certain country reports that it has a higher rate of male births with probability of a boy is 0.6. Assuming the births are random, (i) draw a tree diagram to repersent the distribution of children in families with three children; (ii) find the probability that there are (a) at most one boy and (b) at least one boy in a family of three children. Solution (i): Probability Trees or Tree Diagrams can be used to visualize the events and to calculate simple probabilities. Example: Draw a suitable tree diagram for the experiment of tossing a fair coin two times. Hence list the sample space. Exercise: Draw a tree diagram for the experiment of tossing a fair coin three times. Tree diagram for the distribution of gender of three children 0.6✏✏ ✶B ✏✏ ✑ B PPP ✸ 0.6✑ qG 0.4 P ✑ ✑ 0.6 ✏ ✶B B PP ✏✏ P ✑ ✏ 0.6✑✸ q P 0.4 G PP ✑ qG ✑ 0.4PP ◗ 0.6 ✏ ✶B ◗ ✏✏ ✏ 0.4◗ B 0.6✏✏ ✶ s ◗ PP Pq G G ✏✏ 0.4 P ◗ ◗ 0.6✏✏ ✶B 0.4◗ s G ✏✏ ◗ PP Pq G 0.4 P P (BBB) = 0.6 × 0.6 × 0.6 P (BBG) = 0.6 × 0.6 × 0.4 P (BGB) = 0.6 × 0.4 × 0.6 P (BGG) = 0.6 × 0.4 × 0.4 P (GBB) = 0.4 × 0.6 × 0.6 P (GBG) = 0.4 × 0.6 × 0.4 P (GGB) = 0.4 × 0.4 × 0.6 P (GGG) = 0.4 × 0.4 × 0.4 Solution (ii): (a) P (at most 1 boy) = = (b) P (at least 1 boy) = = SydU MATH1015 (2015) First semester Week 2/3 23 P (0 boy) + P (1 boy) ..... 1 − P (3 girls) ..... SydU MATH1015 (2015) First semester 24

Download PDF

- Similar pages
- McLean HB16 Series Instruction manual
- RoboClaw 2x15A motor controller datasheet - Pololu
- HB16 Instruction Manual
- An Introduction to R - The Comprehensive R Archive Network
- An Introduction to R
- Running R on Windows and Macs (Source: Stanford University Social Science Data and Software)
- Aktueller Test der ZERO ONE in der HIFI Test 5/15
- INSTRUÇÃO SUPLEMENTAR - IS
- PCXMC 2.0
- WHITE LIGHTNINGTM WHITE LIGHTNINGTM
- Bivariate Data in R: Scatterplots, Correlation and Regression
- KLC-103B
- SPSS Basics
- User Manual - Smart Teachers
- datasheet for TCM1512B by Texas Instruments