What is hypothesis testing?

The video covered three crucial concepts about hypothesis testing. These include:

• Why should we perform hypothesis testing?
• What is hypothesis testing?
• What are null and alternative hypotheses?

It is important to note that a hypothesis might be formed using the sample data set while a hypothesis needs to be tested on the entire population that you want to draw conclusions about. Hence, as a part of hypothesis testing, data scientists must collect all the required data about the entire population.

Setting up a null hypothesis

In the earlier activity, you created a hypothesis for your business problem. In order to test a hypothesis, you will also need to frame a null hypothesis because the decision in a hypothesis test is to retain or reject the null hypothesis. So, let’s learn how to create a null hypothesis.

The null hypothesis usually defines a status quo (meaning nothing has changed) and, as a data scientist, you are always trying to find evidence against the status quo. The null hypothesis is assumed to be true before we collect data.

Let’s consider an entrepreneur who may be interested in whether his product sales are impacted by the climate in that location. To test this, the entrepreneur collects his sales data, climatic conditions, and other variables to study the interaction. The null hypothesis, in this case, would be that the sales data remains unaffected by the climatic conditions.

Next, let’s learn about the steps in hypothesis testing through a video.