Updated:
Based on the machine learning canvas, I wrote down a hypothesis test canvas to clarify what hypothesis we are testing, what data is employed and how the results of test will be used.
Context | Value Proposition | Data Sources | |
---|---|---|---|
Idea | Who will use / be affect by the conclusion of test? | What are we trying to test? | Where can we get data from? |
A: | A: | A: | |
Specs | Hypothesis | Evaluation | Dataset |
What is the null hypothesis? | How we verify the quality of data used to verify the hypothesis? | How do we build the table? | |
A: | A: | A: | |
What is the alternative hypothesis? | Who has knowledge about the data? | ||
A: | A: | ||
What is the statistical test employed? (eg t test) | Columns (features) used | ||
A: | A: | ||
What are the statistical assumptions? (eg independence, distributions etc) | |||
A: | |||
What is the significance level? (eg 5%) | |||
A: | |||
What is the statistical power level? (eg 80%) | |||
A: | |||
Ops | Using the results | When do we rerun the test | |
When to we use the test results? | When to we need to update the data and rerun the test? | ||
A: | A: | ||
How do we use the results and confidence values? | |||
A: |
Comments