SLIDE 1
Confidence intervals
P R AC TIC IN G STATISTIC S IN TE R VIE W QU E STION S IN P YTH ON
Conor Dewey
Data Scientist, Squarespace
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Intro to sampling
Wikimedia
1
Confidence inter v als P R AC TIC IN G STATISTIC S IN TE R VIE W - - PowerPoint PPT Presentation
Confidence inter v als P R AC TIC IN G STATISTIC S IN TE R VIE W - - PowerPoint PPT Presentation
Confidence inter v als P R AC TIC IN G STATISTIC S IN TE R VIE W QU E STION S IN P YTH ON Conor De w e y Data Scientist , Sq u arespace Intro to sampling 1 Wikimedia PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON What is a confidence
SLIDE 2
SLIDE 3
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
What is a confidence interval?
Wikimedia
1
SLIDE 4
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Calculating confidence intervals
SLIDE 5
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Calculating confidence intervals
SLIDE 6
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Example: means
import scipy.stats as st a = range(10,14) st.t.interval(0.95, len(a) - 1, loc = np.mean(a), scale = st.sem(a)) (9.446, 13.554)
SLIDE 7
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Example: proportions
from sm.stats.proportion import proportion_conf proportion_confint(4, 10, .05) (0.0964, 0.7036)
SLIDE 8
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Summary
Sampling Condence intervals Example
SLIDE 9
Let's prepare for the interview!
P R AC TIC IN G STATISTIC S IN TE R VIE W QU E STION S IN P YTH ON
SLIDE 10
Hypothesis testing
P R AC TIC IN G STATISTIC S IN TE R VIE W QU E STION S IN P YTH ON
Conor Dewey
Data Scientist, Squarespace
SLIDE 11
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Quick review
xkcd
1
SLIDE 12
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Assumptions
Random sampling Independent observations Normally distributed Constant variance
SLIDE 13
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Generating hypotheses
SLIDE 14
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Which test to use
SLIDE 15
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Evaluating results
0,4 0,4 0,3 0,3 0,2 0,2 0,1 0,1
- 4
- 4
- 3
- 3
- 2
- 2
- 1
- 1
1 2 3 4
x
Wikimedia
1
SLIDE 16
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Types of errors
AB Tasty
1
SLIDE 17
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Summary
Quick review Assumptions Testing process Types of errors
SLIDE 18
Let's prepare for the interview!
P R AC TIC IN G STATISTIC S IN TE R VIE W QU E STION S IN P YTH ON
SLIDE 19
Power and sample size
P R AC TIC IN G STATISTIC S IN TE R VIE W QU E STION S IN P YTH ON
Conor Dewey
Data Scientist, Squarespace
SLIDE 20
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Power analysis
Many Few
Public domain vectors
1
SLIDE 21
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Moving parts
Eect size Signicance level Power Sample size
SLIDE 22
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Calculating sample size
zt_ind_solve_power() tt_ind_solve_power() proportion_effectsize()
SLIDE 23
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Example: conversion rates
from statsmodels.stats.power import zt_ind_solve_power import statsmodels.stats.proportion as prop std_effect = prop.proportion_effectsize(.20, .25) zt_ind_solve_power(effect_size=std_effect, nobs1=None, alpha=.05, power=.80) 1091.8962
SLIDE 24
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Example: conversion rates
from statsmodels.stats.power import zt_ind_solve_power import statsmodels.stats.proportion as prop std_effect = prop.proportion_effectsize(.20, .25) zt_ind_solve_power(effect_size=std_effect, nobs1=None, alpha=.05, power=.95) 1807.76215
SLIDE 25
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Summary
Power analysis Moving parts Example
SLIDE 26
Let's prepare for the interview!
P R AC TIC IN G STATISTIC S IN TE R VIE W QU E STION S IN P YTH ON
SLIDE 27
Multiple testing
P R AC TIC IN G STATISTIC S IN TE R VIE W QU E STION S IN P YTH ON
Conor Dewey
Data Scientist, Squarespace
SLIDE 28
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Multiple comparisons problem
xkcd
1
SLIDE 29
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Correcting for multiple comparisons
GraphPad
1
SLIDE 30
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Common approaches
Bonferroni correction Sidak correction Step-based procedures Tukey’s procedure Dunnet’s correction
SLIDE 31
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Bonferroni correction
Wikimedia
1
SLIDE 32
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Example
from statsmodels.sandbox.stats.multicomp import multipletests p_adjusted = multipletests(pvals, alpha=.05, method='bonferroni') print(p_adjusted[0]) print(p_adjusted[1]) [ True False False False False] [0.05 0.25 0.5 1. 1. ]
SLIDE 33
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Side effects
What’s wrong with Bonferroni adjustments
1
SLIDE 34
PRACTICING STATISTICS INTERVIEW QUESTIONS IN PYTHON
Summary
Multiple comparisons problem Common correction approaches Bonferroni correction
SLIDE 35
Let's prepare for the interview!
P R AC TIC IN G STATISTIC S IN TE R VIE W QU E STION S IN P YTH ON