A/B tests are very commonly performed by data analysts and data scientists. It is important that you get some practice working with the difficulties of these. For this project, you will understand the results of an A/B test run by an e-commerce website and helping the company understand through statistical conclusions, if they should implement the new page, keep the old page, or perhaps run the experiment longer to make their decision.
practice working on A/B testing projects and their practical difficulties perform A/B tests and make recommendations backed by computed inferences
The Project was divided into Three Parts i.e. :
- Perform data cleaning on our dataset.
- Extract some statistics from the cleaned dataset to gain an overall picture on the test result.
- Use hypothesis testing to validate our assumption regarding conversion rate of two landing pages.
- Utilize another method with statsmodels library and compare the result to our previous findings to see if there are any discrepancies.
- Perform logistics regression and then analyze our hypotheses and outputs, especially with regard to our conclusion in Part 2.
- Provide further analysis by adding more variables to our model while taking into account the advantages and disadvantages of this decision.
Based on the statistical tests I used, the Z-test, logistic regression model, and actual difference observed, the results have shown that the new and old page have an approximately equal chance of converting users. We fail to reject the null hypothesis. I recommend to the e-commerce company to keep the old page. This will save time and money on creating a new page.