Top 10 Things to Know for the Exam

BM 1200 #1
Quantitative Methods & Analytics

Week 11:
Review
Top 10 Things to Know for the Exam
Where and When the Exam is!!!

Thank you for taking this class!

Best wishes for the future!

Announcements Top 10 Things to Know for the Exam

} Exam
}

}
The final exam is currently scheduled for Thursday, 14 January
2021 at 9:00am
The duration of the exam will be 2 hours 15 minutes.
#2
} Please follow course office announcements for more information
on how it will be administered.
} Closed books, Lecture notes, Tutorial Solutions, etc.
} An Appendix allowed with the exam will contain the main
formulas, the z-table, and the t-table.
Data Types
} Please make sure to have your calculators ready.
} Best way to prepare is to make sure you are comfortable with
all the material in the lecture slides, the examples we did in
class, and the tutorial questions.
} Practice Exam Questions on Moodle with its solutions.
4
Data Types: Categorical and Numerical Data Top 10 Things to Know for the Final

Data
#3
Qualitative Quantitative
(Categorical) (Numerical)

Examples: Graphical/Numerical Descriptions of

n Marital Status Discrete Continuous
n
n
Political Party
Eye Color
Examples: Examples:
Data
(Defined categories)
n Number of Children n Weight
n Defects per hour n Voltage
(Counted items) (Measured
characteristics)

5 6

Scatter Diagram Example – Fertility Rate Bar/Pie/Line Chart Example – CO2 Emissions

Source: Google public data Source: europa.eu

7 8
 Bar/Pie/Line Chart Example – CO2 Emissions Histograms

Histogram
Credit Card Account Balance Histogram
90
80
70
60

Frequency
50
40
30
20
10
0
199.99 399.99 599.99 799.99 999.99 1199.99 1399.99 1599.99
Credit Card Account Balance
Source: europa.eu
9 10

Numerical Descriptions Top 10 Things to Know for the Final

Center and Location

Describing Data Numerically

Other Measures of Variation

#4
Location
Mean Range
Percentiles
Median Interquartile Range
Mode
Quartiles
Variance
Probability Basics
Weighted Mean
Standard Deviation

Coefficient of
Variation

12
Probability Basics Probability Basics

Rule 1: Probability of any } Conditional Probability: Probability of A given B = P(A|B)

} } Additive Rule: For any two events
elementary outcome must fall P(A Ç B)
A and B P(A | B) =
in the interval from zero to one P(B)
P(A or B) = P(A) + P(B) – P(A and B) } Multiplicative Rule:
P(A Ç B) = P(B)P(A | B)
0 ≤ P(ei ) ≤ 1
= P(A)P(B | A)
} Venn Diagrams } Independence: Two events are independent if the occurrence of one event has
} Rule 2: Probabilities for all
no effect on the probability that the other event occurs.
elementary outcomes in an
If A and B are independent, and P(A)>0 and P(B)>0
experiment (the sample space)
must sum to one P(A Ç B)
P(A | B) = = P(A)
P(B) To show independence,

å P (e ) = 1
all elementary
i P(B | A) =
P(A Ç B)
P(A)
= P(B)
sufficient to show one of
these are equivalent
outcomes in S
P(A Ç B) = P(A)P(B)

} Mutually exclusive events are not independent!

13 14

Top 10 Things to Know for the Final Binomial Distribution

} Binomial Distribution:

#5 ! "=$ =
%!
$! % − $ !
(! 1 − ( "#!

}P(x) = probability of x successes

} x = number of successes, (x = 0, 1, 2, ..., n)
} n = random sample size

Discrete Distributions } p = probability of a success

} n! = n(n-1)(n-2)……1
} 0!= 1 (by definition)
In addition,
} the mean of the distribution is given by µX = E[X] = !"
} the standard deviation is given by !"(1 − ")
15 16
Poisson Distribution Top 10 Things to Know for the Final

} Poisson Distribution:

! "=$ =
*+ !

$!
, #$% #6
} x = number of occurrences in segment of interest
} l = average rate of occurrence per unit segment (e.g., # of calls per
hour)
} t = size of the segment of interest (e.g., a three-hour interval) Continuous Distributions
} e = base of the natural logarithm system (2.71828...)
In addition,
} the mean of the distribution is given by µX = E[X] = '(
} the standard deviation is given by '(

17 18

Uniform distribution Normal distribution z-Table: Given z, Find Probability

The time spent by customers entering an Apple store is uniformly

distributed between 5 and 25 minutes.
1/20 Z ~ N(0,1)
f(x)

5 19 25 x
1. What is the probability that a customer spends less than 8 min? Z 0 1 Z
P(X < 8) = (8 - 5) * 1/20 = 3 / 20 = 0.15
2. What is the probability that a customer spends more than 19
} P(Z <=1) = 0.841
min?
P(X > 19) = (25 – 19) * 1/20 = 6/20 = 0.3

19 20
Normal distribution: Comparing x and z units Top 10 Things to Know for the Final

μ=6
σ=2 X~N(µ=6, s=2)
Z~N(µ=0, s=1)
#7
6 9 X
0 1.5 Z
The distribution is the same, the scale has changed. The properties of the distributions of
In original units (x) or in standard deviations (z).
x−µ Ex: μ = 6, σ = 2, x = 0 => z = (6 – 6) / 2 = 0 samples collected from a population
(X to Z) z=
σ μ = 6, σ = 2, x = 9 => z = (9 – 6) / 2 = 1.5

x = µ + zσ Ex: μ = 6, σ = 2, z = 0 => x = 6 + 2*0 = 6

(Z to X)
μ = 6, σ = 2, z = 1.5 => x = 6 +1.5*2 = 9
21 22

Sampling distributions Top 10 Things to Know for the Final

} Means Standard error of x!
}

}
Sampling distribution of sample means has:

regardless of n
µx = µ X s x = s X
n
When population is normally distributed, distribution of sample means is normal
#8
} Central Limit Theorem: When population not normal, if sample size n ≥ 30, distribution
of sample means is normal:
s
x ~ N (µ, ) Standard error of p"
n
} Proportions
} Sampling distribution of sample proportions has µ pˆ = p s pˆ =
p(1 - p)
n
Confidence Intervals
} Central Limit Theorem: If sample size is such that np > 5 and n(1-p) > 5, distribution of
sample means is normal:
p(1 - p)
pˆ ~ N ( p, )
n

23 24
 Confidence intervals Key learning points: Confidence intervals

The general formula for all confidence intervals with confidence Mean, σ known
level 100(1- a) % is: • n ≥ 30 if population is not normally s
distributed
x ± z(a / 2)
Point estimate from sample: x! or p"
n

Point Estimate ± (Critical Value)(Standard Error) Mean, σ unknown

s
•
•
s = sample standard deviation
t has n-1 degrees of freedom
x ± t(a / 2 )
Critical z (or t)-value: determined by the confidence
coefficient, a
Standard error of sampling
• n ≥ 30 if population not normally n
distribution of x! or p"
distributed

Proportion
Confidence level:
• p̂ = sample proportion pˆ (1 - pˆ )
1. Specifies a • np>5; n(1-p)>5 pˆ ± z(a / 2)
2. 100(1-a)% is the percentage of these intervals that will n
contain the unknown population parameter

25 26

Top 10 Things to Know for the Final Hypothesis Testing Key Steps
1. Identify the appropriate hypothesis test type:
e.g., mean, proportion

#9
}

2. Construct the null and alternate hypothesis

} The alternate hypothesis is what we are trying to show.

3. Use sample data to compute sample statistic

} e.g., for example, sample mean or sample proportion

4. Compute corresponding test statistic (z or t)

Constructing Hypothesis
Obtaining the p-value 5. Compute p-value for test statistic.
} The probability of getting that test statistic or one more extreme (in the
Deciding whether to reject the Null Hypothesis direction of the alternative hypothesis) if the null is true

6. Compare p-value to the significance level, !

} Reject Ho if p-value is < ! (met the burden of proof)
} Do not reject Ho: p-value is > ! (didn’t meet the burden of proof)
27 28
 Recap: Hypothesis Testing – Calculating test statistic Always compute p-value in direction of alternative hyp.
Test Statistics for Population Parameters (Large samples) (1) Lower tail test extreme
H0: μ ≥ μ0
Parameter Test Statistic Standard Error HA: μ < μ0 P(Z < z)
or P(T < t)
Population mean, µ: σ Known x − µ0 σx =
σ 0
z=
(large sample: n ≥ 30)
σx n (2) Upper tail test
extreme
x - µ0 s H0: μ ≤ μ0
Population mean, µ: σ Unknown sx = x
t= HA: μ > μ0 P(Z > z)
(large sample: n ≥ 30)
sx n
or P(T > t)
Population proportion, p pˆ - p0 p0 (1 - p0 ) 0
z= s pˆ =
(3) Two-tailed test
(large sample: p*n > 5 and (1-p)*n > 5)
s pˆ n
extreme
extreme
H0: μ = μ0
HA: μ ≠ μ0

0
2*P(Z > |z|) or 2*P(T > |t|)
29 30

Top 10 Things to Know for the Final Reading the Regression Output Table

} Excel Output:

# 10 Regression Statistics
Multiple R
R Square
0.76211
0.58082
The regression equation is:

Adjusted R Square 0.52842

house price = 98.24833 + 0.10977 (square feet)
Standard Error 41.33032
Reading the Regression Output Table Observations 10

ANOVA
df SS MS F Significance F
Regression 1 18934.9348 18934.9348 11.0848 0.01039
Residual 8 13665.5652 1708.1957
Total 9 32600.5000

Upper
Coefficients Standard Error t Stat P-value Lower 95% 95%
Intercept 98.24833 58.03348 1.69296 0.12892 -35.57720 232.07386
31 Square Feet 0.10977 0.03297 3.32938 0.01039 0.03374 0.18580
Reading the Regression Output Table Reading the Regression Output Table
This is the p-value: 0.01039 } Excel Output:
Regression Statistics Since 0.01039 < 0.05, regression model is
Multiple R 0.76211 significant at the 0.05 significance level.
Regression Statistics
R Square 0.58082
Multiple R 0.76211
Adjusted R Square 0.52842
R Square 0.58082
Standard Error 41.33032
Observations 10
Adjusted R Square 0.52842 58.08% of the variation in
Standard Error 41.33032 house prices is explained by
ANOVA Observations 10 variation in square feet
df SS MS F Significance F
Regression 1 18934.9348 18934.9348 11.0848 0.01039
ANOVA
Residual 8 13665.5652 1708.1957 df SS MS F Significance F
Total 9 32600.5000 Regression 1 18934.9348 18934.9348 11.0848 0.01039
Residual 8 13665.5652 1708.1957
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Total 9 32600.5000
Intercept 98.24833 58.03348 1.69296 0.12892 -35.57720 232.07386
Square Feet 0.10977 0.03297 3.32938 0.01039 0.03374 0.18580
Upper
Coefficients Standard Error t Stat P-value Lower 95% 95%
Intercept 98.24833 58.03348 1.69296 0.12892 -35.57720 232.07386
33 Square Feet 0.10977 0.03297 3.32938 0.01039 0.03374 0.18580

Tools for making business decisions

Decision Making Toolbox

Inference/Predictions:
Tools for dealing with uncertainty
Confidence Intervals
Probability
Hypothesis testing
Discrete and Continuous Distributions
Regression

BM1200
 What have we accomplished? What have we accomplished?

} Probability, Random Variables } Sampling, Confidence intervals, Hypothesis testing

} The language of uncertainty } Understanding samples, surveys, opinion polls
} Using additional information to revise/update probability of an event } Providing estimates for unknown quantities using samples
} Testing/supporting/refuting claims based on statistical evidence

Operations
Modeling Consumer Behavior Marketing
Planning for Demand Uncertainties R&D, etc.
Service Operations, customer wait times

Source: energysavingtrust.org.uk

If you work for an online retailer, now you know how to use the consumer behavior
information to update purchase probabilities. If you work for a policy organization, you can help identify whether online retail or
high street shopping is more energy efficient.

What have we accomplished? Thank you - Online Course Evaluations

} Regression
} Constructing models based on sample data, to:
} Thank you so much for your feedback
} Infer degree of influence of one factor on another
} I really, really value your input
} Make predictions about unknowns using regression model
} Your evaluations will encourage me
} Your comments will help me refine my teaching and this module

Source: McKinsey&Co

If you work for a consulting company, now you know how to use regression in
estimating the influencers of consumer purchase decisions.
40
Thank you!

Congratulations for your hard work throughout

the term!

I really enjoyed teaching this module

and having you in class!

I hope you enjoyed the class as well!

Best wishes to all of you…