Master Statistics for Data Analysis

Describe a real dataset with statistics

1 week. The computation takes a day. The interpretation takes the rest of the week — most people underestimate how hard it is to say what numbers mean rather than just what they are.

Run and interpret a hypothesis test

1–2 weeks. Forming the hypothesis takes longer than running the test. Most people want to run tests and then find hypotheses that match — resist this; the hypothesis must come first or the test is meaningless.

Build and evaluate a regression model

1–2 weeks. Building the model takes a day. Understanding why the overfit model is worse — genuinely understanding it, not just observing it — takes the rest of the time.

Design and analyse an A/B test

2 weeks. The power calculation takes a day to understand and implement. The product brief takes longer than the statistics — translating statistical output into business language is a distinct skill that most statistics courses never teach.

Communicate findings to a non-technical audience

1 week. The statistics are done. The communication is the hard part. Most people who are good at analysis are poor at presentation — not because they lack skill, but because they forget what it is like not to know what they know.