You are on page 1of 6

Please List your questions below:

can you summarize and share when and how to to use Excel formulae eg: NORM.DIST, NORM.S.INV etc., also python code e
All have been explained as silo's hence confused
Type 1 and type 2 clarification
1. Don't we ever use t-test on a samle less than 30?
2. Usually, population variance is not known unless it is census or an entirely population study. How to deal with this scenario?
3. Is it possible to exemplify binomial, Poisson and normal distribution with a single example, though normal is continuous prob
4. Kindly explain, which type of error is more dangerous, type 1 or type 2 with an example.
5. Also explain right or left, positive and negative skewness with example. (How do we determine the sides)
1.) Please show Some examples of back calculating type 1 error, type 2 error and power of the test from problem statements.
2.) Some examples of how the null and hypothesis setup changes: when a company is doing the test themselves and when a
reference to one of the problem statement in Week 3 SMDM project.
1. Example on Bayes theorem
2. Inferential statistics
3. How to write a code in simple manner for statistics function
How to find the null hypothesis when the question give range of two number?
Any easy way to determine H0 and H1?
Clarify on application of different type of distribution
Clarity on z value
Most of the videos shown how to use excel for calculating the statistical information, can you explain the same using the Python
how to generate sub total tables according to two or more columns for a given column and also the row total for that
1. Please explain Conditional probability with examples
2. Please elaborately explain about the Bayer's theorem with the hands on example
If stats.norm.pdf already gives us the area under the curve at a point which is also equal to the probability value at a point then
point.
To find out if there is any outlier in the dataset, should we calculate it for each and every variable?
How to calculate conditional probability? It is a bit confusing
On Mcdonalds assignment it is asked which variable has outliers, what is the variable in this data set?

1) Ho:X1<=X2
H1:X1>X2
By reject the null hypothesis, we prove that X1 is not equal to X2 and assume that X1>X2 but may be X1 is less than X2, so ho
2) t critical and z critical are same ?
3) can we spot p value on a normal distribution graph (how can we visualise p value)
4) different between p_value = stats.norm.sf(abs(z)) and p_val = stats.norm.cdf(z)
H0 Null Status Quo Assumed to be TRUE
Ha Alternate Claim/Assertion What you want to Prove

Court Trial
H0 Person is innocent until proven guilty
Ha Person is guilty

Confusion Matrix Outcome


Actual NG G Incorrect Rejection of Null Hy
NG TN FP Type-1 Error
G FN TP
1 POS
0 NEG Typ2 Error
Beta 0.1 or 10%
1-beta 0.9 (Power)

Sample Size

p-value > alpha If P is high, null will fly


p-value < alpha If P is low, null will go

Type of Tests

# Samples Condition
1 Sample POP SD is known

1 Sample POP SD is unknown

H0: Data is normal H0 `=250


Ha: Data is not normal Ha <250
Hyp Mean 250
shapiro 1. Sample should be normally distributed Sam Mean 248.5
levene 2. Samples should have equal variance Sam SD 4.8
Sam Size 36
DOF 35

tstat -1.875 -1.689572


p-value 0.034578
Sample
2 Samples Size Independent
2 Samples shold be Dependent
shold be
equal Before/After

2+ Samples

P(A|B) Mutually Exclusive


Independent

Addition Rule

Limit the Sample Space P(AuB) P(A)+P(B) - P(AnB)

Joint Prob Multiplication


P(AnB) P(A)*P(B) Indepenent

Bayes Rule
P(A|B) P(B|A) * P(A) / P(B)

Assume 10,000 components

P(A) 0.7 A 7000


P(B) 0.18
P( C) 0.12 B 1800

P(F|A) 0.02 C 1200


P(F|B) 0.1
P(F|C) 0.18

P(F) 0.0536

P(B|F) P(F|B) * P(B) / P(F)

0.335821

Bag1 4R, 6W
Bag2 8R, 2W
Bag3 3R, 6W

If I pick a Red ball at random, what is the Prob it comes from Bag2?
0.533333

$10,000
A B A B
15 150 -1.180157 -1.180157
20 200 -0.833052 -0.833052
35 350 0.208263 0.208263
40 400 0.555368 0.555368
50 500 1.249578 1.249578

mean 32 320
sd 14.40486 144.0486
cv 0.450152 0.450152
Assumed to be TRUE
What you want to Prove >, <, !=

Incorrect Rejection of Null Hypothesis


Type-1 Error Alpha
Level of Significance 0.05 or 5%

Confidence Level
1-alpha 0.95 or 95%

h, null will fly Fail to reject H0


, null will go Reject H0

Type of Test
z Test z = x-mu/SE SE = SD/sqrt(n)
norm.s.dist()
stats.norm.cdf()

T Test tstat = x-mu/SE SE = SD/sqrt(n)


t.dist()
DOF = n-1

3 5 9
10 5 25

A B
12 11
17 15
2 1
16 7
0 6
2 Sample T Test 0 9
Paired T Test 6 17
6 3
14 14
9 4
ANOVA 2 13
F-Test 0 NaN
6 NaN
14 NaN

0,000 components

S 6860
F 140
S 1620
F 180
S 984
F 216

180
total failuers 536 0.335821

You might also like