Congratulations! You now know how to run a binomial hypothesis test using a SciPy function — or by simulating it yourself! This will serve you well as a data scientist because it will enable you to investigate what’s going on if pre-written functions return surprising results. You also now have a conceptual understanding of how a binomial test works and what questions it aims to answer. To summarize, here are some of the things you’ve learned about hypothesis tests in general:
All hypothesis tests start with a null and alternative hypothesis
Outcomes of a hypothesis test that might be reported include:
- confidence intervals
- p-values
A hypothesis test can be simulated by:
- taking repeated random samples where the null hypothesis is assumed to be true
- using those simulated samples to generate a null distribution
- comparing an observed sample statistic to that null distribution
Instructions
As a final exercise, the solution code for the previous exercise is available to you in script.py. As a challenge, see if you can re-write the simulation-based binomial test function so that it has an input named alternative_hypothesis
that can be equal to 'less'
, 'not_equal'
, or 'greater'
. Then change the function so that it performs the appropriate one- or two-sided test for the alternative hypothesis provided. Solution code is available to you in solution.py if interested.