Download Datasets and Presentation slides for this post HERE

Hypothesis testing lets you answer questions about your datasets in a statistically rigorous way. In this course, you'll grow your Python analytical skills as you learn how and when to use common tests like t-tests, proportion tests, and chi-square tests. Working with real-world data, including Stack Overflow user feedback, you'll gain a deep understanding of how these tests work, and the key assumptions that underpin them. You'll also discover how different tests are related using the “there is only one test" framework, before learning how to use non-parametric tests to go beyond the limitations of side-step the requirements of hypothesis tests.

import pandas as pd
import numpy as np
import warnings
import matplotlib.pyplot as plt
plt.rcParams['figure.figsize'] = [8, 6]

pd.set_option('display.expand_frame_repr', False)

warnings.filterwarnings("ignore", category=DeprecationWarning)
warnings.filterwarnings("ignore", category=FutureWarning)

Yum, That Dish Tests Good

How does hypothesis testing work and what problems can it solve? To find out, you’ll walk through the workflow for a one sample proportion test. In doing so, you'll encounter important concepts like z-scores, p-values, and false negative and false positive errors.

To the lab for testing

Uses of A/B testing Calculating the sample mean Calculating a z-score

A tail of two z's

Criminal trials and hypothesis tests Left tail, right tail, two tails Calculating p-values

Statistically significant other

Decisions from p-values Calculating a confidence interval Type I and type II errors

Pass Me ANOVA Glass of Iced t

In this chapter, you’ll learn how to test for differences in means between two groups using t-tests and extend this to more than two groups using ANOVA and pairwise t-tests.

Is this some kind of test statistic?

Hypothesis testing workflow Two sample mean test statistic

Time for t

Why is t needed? The t-distribution From t to p

Pairing is caring

Is pairing needed? Visualizing the difference Using ttest()

P-hacked to pieces

Visualizing many categories ANOVA Pairwise t-tests

Letting the Categoricals Out of the Bag

Now it’s time to test for differences in proportions between two groups using proportion tests. Through hands-on exercises, you’ll extend your proportion tests to more than two groups with chi-square independence tests, and return to the one sample case with chi-square goodness of fit tests.

Difference strokes for proportions, folks

t for proportions? Test for single proportions

A sense of proportion

Test for two proportions proportions_ztest() for two samples

Declaration of independence

The chi-square distribution How many tails for chi-square tests? Chi-square test of independence

Does this dress make my fit look good?

Visualizing goodness of fit Chi-square test of goodness of fit

Time to Define the Relationship

Finally, it’s time to learn about the assumptions made by parametric hypothesis tests, and see how non-parametric tests can be used when those assumptions aren't met.

What do you assume?

Common assumptions of hypothesis tests Testing sample size

Assumptions not met

Which parametric test? Wilcoxon signed-rank test

Look ma! Still no parameters!

Wilcoxon-Mann-Whitney Kruskal-Wallis

Congratulations!