|
QMM 510 Fall 2014 Stats Analysis for Managers |
Detailed Syllabus Updated
Nov 18, 2013 |
Prof. Doane doane@oakland.edu |
|||||
|
Week |
Topics and |
Due |
Reading |
|
||
|
Week
1 Sep 1-5 |
·
Getting
started: o
self-introductions o
course
format, syllabus, projects o
goals:
short run vs long run ·
Resources
available: o
textbook,
e-book o
OLC
(http://www.mhhe.com/doane4e) o
Connect
(http://connect.mcgraw-hill.com/class/qmm510fall2014) o
Moodle
(https://moodle.oakland.edu/) o
MegaStat
(http://www.mhhe.com/megastat) o
LearningStats
(http://www.mhhe.com/doane4e) o
Doane
(http://www.sba.oakland.edu/faculty/doane) ·
Challenges
for MBAs: o
big
data, powerful tools o
ethical
guidelines o
the
ideal statistician o
when
to hire a consultant o
critical
thinking ·
Collecting
data: o
data
types and measurement o
random
sampling (4 methods) o
nonrandom
sampling (3 methods) o
randomizing
a data column in Excel o
sources
of error, survey types o
response
scales (e.g.,Likert) ·
Describing
data visually: o
center,
variability, shape o
stem-and-leaf,
dot plots, histograms o
frequency
polygon, ogive (MegaStat) o
examples:
birth weight, CEO salaries o
scatter
plots (Excel, MegaStat) ·
Assignments:
o
Connect
C-1 (covers chapters 2–3) o
Project
P-1 (data, tasks, questions) |
None |
Ch 1 (read all but only for general
information) Ch 2 (read all, focusing on data types
and random sampling) Ch 3 (read all, but focus on section 3.2
on frequency distributions and histograms) |
|
||
|
Week |
Topics and |
Due |
Reading |
|
Week
2 Sep 8-12 |
·
Center
and variability: o
center
(mean, median, mode, etc) o
variability
(variance, std dev, etc) o
Excel
functions (Appendix J) o
example:
birth weight ·
Standardized
data: o
sorting,
standardizing, z-scores o
normal
distribution as a benchmark o
Empirical
Rule (MegaStat) o
outliers
and unusual observations o
Excel
functions (Appendix J) o
examples:
birth weight, voting o
using
MegaStat and Excel ·
Quantiles:
o
percentiles,
quartiles, boxplots o
fences,
another view of outliers o
examples:
city MPG, birth weight ·
Correlation,
grouped data, shape: o
scatter
plots o
correlation
coefficient o
covariance
– population, sample o
mean
from grouped mean o
skewness,
kurtosis (Excel) ·
Assignments:
o
Connect
C-2 (covers chapter 4) o
Project
P-1 (data, tasks, questions) |
C-1 covering
chapters 2-3 (due at 11:59 pm on Monday Sep 8) |
Ch 4 sections 4.1 thru 4.3) Ch 4 (section 4.4) Ch 4 (section4.5) Ch 4 (sections 4.6-4.8) |
|
Week |
Topics and |
Due |
Reading |
|
Week
3 Sep 15-19 |
·
Time
series trends: o
time
patterns (visual) o
trend
fitting (3 models) ·
Trend
forecasting: o
assessing
fit (4 criteria) o
trend
forecasting o
example:
Amazon sales ·
Forecasts
with seasonality: o
seasonal
factors o
seasonally
adjusted forecasts o
MegaStat
seasonal o
example:
Amazon sales ·
Moving
averages: o
moving
averages (if no trend) o
exponential
smoothing (brief) ·
Assignments:
o
Project
P-2 (data, tasks, questions) |
C-2 covering
chapter 4 (due at
11:59 pm on Monday Sep 15) |
Ch 14 (sections 14.1 and 14.2) Ch 14 (section 14.3)
Ch 14 (section 14.4) |
|
Week
4 Sep 22-26 |
·
Probability:
o
definitions
of probability o
independent
events o
contingency
tables o
counting
(review as needed) ·
Discrete
Probability Distributions: o
what
is a probability distribution? o
expected
value: E(X) and V(X) o
definitions
of pdf and cdf o
uniform
and Excel’s =RANDBETWEEN ·
Binomial
distribution: o Bernoulli
events (X = 0, 1) o two
parameters (n, π) o pmf vs
cdf (formulas, Excel) o Excel’s =BINOM.DIST o Excel’s =BINOM.INV o recognizing
a binomial ·
Poisson
distribution: o one
parameter (λ) o pmf vs
cdf (formulas, Excel) o Excel’s =POISSON.DIST o recognizing
a Poisson ·
Other
Discrete Distributions: o hypergeometric
(parameters N, n, s) o geometric
(parameter π) |
P-1 covering
chapters 3-4 (due at
11:59 pm on Mon Sep 22) |
Ch 5 (omit section 5.7) Ch 6 (sections 6.1 and 6.2) Ch 6
(sections 6.3 and 6.4) Ch 6
(section 6.5) |
|
Week |
Topics and |
Due |
Reading |
|
Week
5 Sep 29-Oct 3 |
·
Normal
distribution: o
pdf
vs cdf (formulas, Excel) o
calculating
a standardized z score o
area
for given z using Appendix C o
area
for given x or z using Excel functions §
=NORM.DIST(x, μ, σ, 1) §
=NORM.S.DIST(z,1) ·
Inverse
normal distribution: o
z for given area using Appendix C o
x or z for given area α using Excel function §
=NORM.INV(α, μ, σ) §
=NORM.S.INV(α) o
z values for common areas (.10, .05, etc) o
normal
approximations – why bother? ·
Exponential
distribution: o
pdf
and cdf (formulas, Excel) o
uses
in waiting time or queueing o
example
of prob for given wait o
example
of inverse (wait for given prob) o
example
for given MTBF ·
Other
continuous distributions: o
triangular
distribution §
pdf
and cdf §
mean
and standard deviation §
uses
in “what if” simulation o
uniform
distribution §
mean
and standard deviation §
Excel’s =RAND() |
C-3 covering
chapters 5-6 (due at 11:59
pm on Monday Sep 29) |
Ch 7 (sections 7.1, 7.3, 7.4, and 7.5) Ch 7 (section 7.6) Ch 7 (sections 7.2 ans 7.7) |
|
Week |
Topics and |
Due |
Reading |
|
Week
6 Oct 6-10 |
·
Sampling
Distributions: o
sample
mean is a random variable (r.v.) o
properties
of estimators o
sampled
items vary but mean varies less o
mean
→ normal for small n if sym
pop o
variance
of mean → 0 as n → ∞ ·
CI for
µ with σ known: o
standard
error of the mean is σ/n1/2 o
which
confidence level (.90, .95, .99)? o
use
z if you know σ (or very large n) o
z for given 1–α
(Appx C or Excel) ·
CI
for µ for σ unknown: o
use
t if unknown σ (typical situation) o
t for given 1–α
(Appx D or Excel) o
quick
rule for 95% CI o
conservative
just to use t o
CI
from MegaStat (or Minitab) ·
Conf
interval for a variance: o
requires
a chi-square table (Appx D) o
less
common (use Minitab or MegaStat) |
C-4 covering
chapter 7 (due at 11:59
pm on Monday Oct 6) |
Ch 8 (sections 8.1 thru 8.3) Chapter 8 (section 8.4) Chapter 8 (section 8.5) Ch 8 (section 8.10) |
|
Week |
Topics and |
Due |
Reading |
|
Week
7 Oct 13-17 |
·
Conf
interval for p: o
sample
proportion p = x/n is a r.v. o
the
CLT applies to a proportion p o
standard
error of p is [p(1–p)/n]1/2 o
use
z for CI if n is “large” o 10 “successes” and 10 “failures” o
use
binomial for CI if n is small
(Minitab) o
poll
margin of error assumes π = .50 o
effect
of finite populations ·
Sample
size for p and µ:
o
it
is conservative to assume π =
.50 o
but
sample may be larger than necessary o
sample
size for µ requires assumed σ o
but
there’s no conservative choice for σ o
you
really need a preliminary sample ·
Hypothesis
tests: o
type
I error, type II error, and power o
tradeoff
between type I and II error |
P-2 covering
chapter 14 (due at
11:59 pm on Mon Oct 13) |
Ch 8 (sections 8.6 and 8.7) Ch 8 (section 8.10) Ch 8 (8.8 and 8.9) Ch 9 (section 9.1) |
|
Week |
Topics and |
Due |
Reading |
|
Week
8 Oct 20-24 |
·
Tests
for µ = µ0 (known σ):
o
formulating
the null hypothesis H0 o
sign
of H1 indicates tail of
test (>, ≠, <) o
level
of significance (.10, .05, .01) o
use
z if you know σ (or very large n) o
critical
values for z (Appx C or Excel) o
we encounter
p-values often o
the
p-value always refers to H0 o
similar
interpretation for all tests o
why
are two-tailed tests common? ·
Tests
for µ = µ0 unknown σ:
o
use
t if unknown σ (typical situation) o
critical
values for t (Appx D or Excel) o
conservative
just to use t o
using
MegaStat (or Minitab) ·
One-sample
z-test for π = π0:
o
use
z as long as n is large enough o
10
“successes” and 10 “failures” o
binomial
if n is small (Minitab) o
using
MegaStat (or Minitab) |
C-5 covering
chapter 8 (due at 11:59 pm on Monday Oct 20 |
Ch 9 (sections 9.2 and 9.3) Ch 9 (section 9.4) Ch 9 (pp. 352-353 are very important) Ch 9 (section 9.5) |
|
Week |
Topics and |
Due |
Reading |
|
Week
9 Oct 27-31 |
·
Independent
sample t-tests: o
Case
1: z-test for µ1 - µ2 = 0 (known
population variances, rarely used) o
Case
2: t-test for µ1 - µ2 = 0 (unknown
variances, pooled variance estimate) o
Case
3: t-test for µ1 - µ2 = 0 (unknown
variances, separate s variance estimates) o
p-values and Excel functions o
example
using Excel t-tests o
example
using MegaStat ·
Paired
sample t test: o
really
a one-sample t-test for µd =
0 o
can
you recognize paired data? ·
z-test for π1
- π2 = 0: o
summarized
data o
when
is normality assumption safe? o
advantages
of MegaStat (or Minitab) o
example
using MegaStat ·
F test for σ12 = σ22 : o
F test: Excel vs MegaStat o
Quick
rule for variances |
None (but try to get started on C-6) |
Ch 10 (omit 10.3 and 10.6) |
|
Week |
Topics and |
Due |
Reading |
|
Week
10 Nov 3-7 |
·
Chi-Square
Tests for Independence: o
Chi-square
distribution o
Contingency
tables revisited o
Hypotheses
and decision rule o
Excel’s
=CHISQ.DIST.RT(α,d.f.) o
MegaStat’s
chi-square test o
Cochran’s
Rule (ej ≥ 5) o
Example:
web survey ·
Chi-Square
Goodness-of-Fit tests: o
Degrees
of freedom d.f. = k−m o
Chi-square
GOF test for normality o
Why
Method 3 is preferred o
Disadvantages
of chi-square GOF ·
ECDF
Tests for GOF: o
Kolmogorov-Smirnov
test o
Anderson-Darling
test o
Probability
plots o
Examples:
Minitab and MegaStat o
Advantages
of ECDF tests |
C-6
coveringchapters 9 and10 (due at 11:59 pm on Monday Nov 3 |
Ch 15 (sections 15.1-15.2) Ch 15 (section 15.5) Ch 15 (section 15.6) |
|
Week |
Topics and |
Due |
Reading |
|
Week
11 Nov 10-14 |
·
Correlation
analysis:- o
significance
tests for r o
uses
of correlation analysis ·
Simple
regression: o
OLS
method, assumptions o
interpreting
a fitted regression o
example
(CarSpecs) o
R2
statistic, ANOVA, standard error(se) o
tests
for significance (t-test, F-test) o
unusual
residuals, outliers o
high
leverage observations ·
Tests
for violations: o
non-normality
(histogram, PP) o
autocorrelation
(residual runs plot, DW) o
heteroscedasticity
(residual plots) o
consequences
of violations o
transformations
that may help o
but
are transformations worth it? ·
Project
P-3 (discuss): o
Why
multiple predictors? o
How
many predictors: Evans’ Rule o
Project
P-3 (preview, data, tasks) o
using
the state database o
work
with partners? o
start
with choice of Y o
choose
your Y to be explained o
choose
predictors X1, X2, … o
ask
instructor about proposed model o
post
questions on Moodle (for peers or instructor) |
C-7
coveringchapter 15 (due at 11:59 pm on Monday Nov 10 |
Ch 12 (section 12.1) Ch 12 (sections 12.2 thru 12.6) Ch 12 (section 12.9) Ch 12 (sections 12.8 and 12.10) |
|
Week
12 Nov 17-21 |
·
Multiple
regression: o
model
building, predictor choice o
how
many predictors (Evans) o
tests
for significance (t-tests, F-test) o
quick
rules for t and F, conf. intervals o
should
you omit poor predictors? o
assessing
fit- R2, R2adj, std error
(se) o
Occam’s
Razor, role of sample size ·
Unusual
observations: o
unusual
residuals o
high
leverage observations o
similar
to simple regression ·
Categorical
predictors: o
using
binary predictors (c–1) o
just
like any other predictor, but ... o
the
residual plots look weird ·
Other
Regression Topics: o
Outliers?
Ill-Conditioned Data? o
Significance
in Large Samples? o
Model
Specification Errors? o
Binary
Response? Stepwise Regression? o
Project
P-3 (checkpoints) §
have
you estimated your model? §
variable
scaling questions §
post
questions on Moodle forum |
C-8 Ch 12 (due at 12:59
pm on Monday Nov 17) |
Ch 13 (sections 13.1 thru 13.3) Ch 13 (section 13.8) Ch 13 (section 13.5) |
|
Week |
Topics and |
Due |
Reading |
||
|
Week
13 Nov 24-28 |
·
Multicollinearity
(MC): o
consequences
of MC o
correlation
matrix (for k predictors only) o
collinearity
vs multicollinearity o
variance
inflation factors (VIFs) o
fit:
compare se with mean of Y o
fit:
plot Yactual against Yfitted o
is some explained SSR better than none? o
tiny
competitive advantage in business may be critical (e.g., data mining) ·
Tests
for violations: o
non-normality
(histogram, PP) o
autocorrelation
(residual runs plot, DW) o
heteroscedasticity
(residual plots) o
mostly
the same as simple regression ·
Assignments: o
Project
P-3 (last-minute instructions) o
post
questions on Moodle o
e-mail
questions to instructor |
None (but get started on project P-3) |
Ch 13 (section 13.7) Ch 13 (sections 13.8 and 13.9) |
||
|
Week
14 Dec 1-5 |
·
Wrap-Up:
No mini-lectures posted o
Submit
project P-3 (note unusual due date) o
Last-minute
e-mails to instructor |
P-3 (due at
5:00 on Wed Dec 3) |
None |
||
|
|
|
|
|||
|
|||||