Mean | =AVERAGE() | |||
---|---|---|---|---|
Weighted Mean |
=SUMPRODUCT(weights, values) / n Where n = SUM(weights) |
|||
Median | =MEDIAN() | |||
Mode | =MODE.SNGL() or =MODE.MULT() | |||
Range | =MAX()-MIN() | |||
Class Width | =range / number of classes | |||
Variance | Sample: | =VAR.S() | ||
Population: | =VAR.P() | |||
Standard Deviation (SD) | ||||
Sample: | =STDEV.S() | |||
Population: | =STDEV.P() | |||
Or,=SQRT(variance) | ||||
Skewness |
=3*(mean - median) / SD |
|||
Coefficient of Variation | ||||
=SD / mean * 100 | ||||
Empirical Rule | 0.68, 0.95, 0.997 |
|||
Chebyshev | = 1 - 1 / SD^2 |
|||
Percentile | i = percentile/100 * n Round up if decimal; average if whole. |
|||
IQR, Outliers | = Q3 - Q1 |
|||
Upper Outlier =Q3 + 1.5 * IQR |
||||
Lower Outlier =Q1 − 1.5 * IQR |
||||
Probability | P(A) P(A or B) |
= A / Total = P(A) + P(B) - P(Both) |
||
P(A and B) P(A and B) P(A | B) P(A | B) |
= P(A) * P(B) = P(A) * P(B|A) = P(A and B) / P(B) = P(A|B)*P(B) P(A|B)P(B)+P(A|Bc)P(Bc) = P(B|A)*P(A) / P(B) |
Independent Dependent Conditional Bayes' Rule |
||
Permutation | =PERMUT(n,r) | |||
Combination | =COMBIN(n,r) | |||
Discrete Distributions | ||||
Expected Value: =SUMPRODUCT(x,P(x)) Variance: =SUM(x^2 * p) - (expected value^2) SD: =SQRT(discrete variance) |
||||
Binomial Discrete Distributions | ||||
|
Mean or Expected Value: =n*p Variance: =n*p*(1 - p) SD=SQRT(variance) Probability: =BINOM.DIST(successes,trials,probability,cum) |
|||
Z Score | =STANDARDIZE(x,mean,SD) =(x - mean) / SD |
Normal Distributions | |||
---|---|---|---|
Probability: =NORM.DIST(x,mean,SD,TRUE) Where TRUE means cumulative |
|||
Z to Probability: =NORM.S.DIST(z score,1) |
|
||
|
Probability to Z: =NORM.S.INV(probability) Probability to X: =NORM.INV(probability,mean,SD) |
|
|
Sampling from Normal Distribution, Mean (Central Limit Theorem) | |||
Standard Error: | =SD / SQRT(n) | ||
Probability: =NORM.DIST(x,mean,SD/SQRT(n),1) |
=NORM.DIST() | ||
Where 1 means cumulative | |||
Sampling from Normal Distribution, Proportion | |||
Validation: n*p ≥ 5 and n*(1-p) ≥ 5 | |||
z = (pS - p) / SQRT(p*(1 - p) / n) | |||
Probability: =NORM.S.DIST(z,1) |
|||
Confidence Intervals, Mean | |||
Alpha = 1 - Confidence Level Margin of Error (E): E =CONFIDENCE.NORM(alpha,SD,n) |
|
||
E =CONFIDENCE.T(alpha,SD,n) (small sample) Confidence Interval: = (mean - E) to (mean + E) |
|||
Confidence Intervals, Proportion | |||
z =NORM.S.INV(confidence level + alpha/2) E =z * SQRT(p * (1-p)/n) Confidence Interval: =(proportion - E) to =(prop + E) |
|||
Sample Size, Mean | |||
z = NORM.S.INV(confidence+alpha/2) n = (z * SD / E)^2 |
|||
Sample Size, Proportion | |||
n = p*(1 − p) * (z / E)^2 | |||
Hypothesis Testing Mean | |||
Decision Rule (D.R.): 1-tail: z =NORM.S.INV(confidence level) 2-tail: z =NORM.S.INV(confidence level + alpha/2) Test Statistic: z = (sample mean - mean) / (SD/SQRT(n)) |
|||
Hypothesis Testing Proportion | |||
Test Statistic: z = (ps - p) / SQRT(p*(1 - p) / n) | |||
Hypothesis Testing Small Sample | |||
D.R.: 1 tail: t =T.INV(alpha,df) D.R.: 2 tail: t =T.INV.2T(alpha,df) Where df = degrees of freedom = n - 1 |
|
||
Hypothesis Testing P-Value | |||
1-tail: z =NORM.DIST(sample mean, population mean, SD / SQRT(n),1) 2-tail: z = NORM.DIST(sample mean, population mean, SD / SQRT(n),1)*2 |
|||
Regression Analysis | |||
y = m*x + b (but in statistics it's written y = a + bx) | |||
In Excel, "Multiple R" = coefficient of correlation |