search

sariabod / codeacademy

intro to data analysis
1 stars 0 forks source link

Summary: Meets Expectations #5

Open Caealana opened 6 years ago

Caealana commented 6 years ago

Overall, nice project! Your code is functionally correct and performs the correct analysis of the data. You use both space and time efficient methods. I would just suggest organizing your code and including comments for easier reading.

Your slide presentation looks great! Your introduction and context explanations are clear and concise. You used one overarching hypothesis. However, this project actually attempts to test 3 separate hypothesis: 1) Does the fitness test have an effect on whether or not someone picks up an application? 2) Does the fitness test have an effect on whether or not someone signs up for a membership? 3) Of those who picked up an application, how many purchased a membership? You should make note of these questions/motivations in your slides.

sariabod commented 6 years ago

@Caealana the story telling is probably where I need the most improvement. The second set of results being not as expected threw me for a loop when it came to dealing with the final result.

Is it normal to test and show multiple hypothesis in a single report? To me, it seemed the main question was about the correlation between the fitness test and signing up. The in between steps are good to analyze, but do they need to be presented if they are not part of the original hypothesis? Would they add confusion?

Do you normally put in things like pval into the final report? I am thinking more people will not understand it and keeping the report simple with normal terminology is a better way to go?

Caealana commented 6 years ago

@sariabod I apologize for the delay in my response! I just want to let you know I have seen your questions and I will be responding in the next couple of days. I am a bit busy with travel at the moment, and I need to meet some deadlines for leaving first round comments for a couple of other projects, but I will get back to you as soon as I can.

Caealana commented 6 years ago

@sariabod What did you expect for the second set of results?

Generally, it is up to personal preference. I usually suggest showing the multiple hypotheses because the results for this project show that the different hypotheses had different results. You can draw specific and different conclusions from each of the hypothesis tests. Here is one example of a project that does so.

For reporting the results from your statistics tests, you should usually include your pval. It is usually quite easy to explain how to interpret it (significant vs not significant) and it is a single, compact number that is easy to show as the result of your test. There is always that difficult balance of trying to keep your report accessible to a wide audience but also making sure to go into enough detail to provide enough support for your conclusions.

sariabod commented 6 years ago

@Caealana Treating them as different hypotheses would have cleared it all up. I was making it one big experiment.