Employee Exit Survey Project

Hello Everyone! I am submitting the guided project on Employee Exit Surveys. In addition to answering the two required questions, I also explored the resignations from various other lenses. This project, by far, demanded the most data cleaning and it took me two weeks to finish this one. :weary:
I request you to have a look and provide your valuable feedback. :smiley:

2 Likes

Hi @sahiba.kaur.stats

Your project looks great with all the charts and detailed analysis. I can see that there is another employee exit project submitted by another community member. Why don’t you go ahead and have a look and give a feedback. I’ll request the other person to do that same as well. Here is the link.
My answer to: Clean and Analyze Employee Exit Surveys

Click here to view the jupyter notebook file in a new tab

1 Like

Hi @sahiba.kaur.stats! Thanks for sharing your project with the Community :slight_smile: I like your detailed conclusions and the plots are also nice. But I mostly liked your recommendations of the steps that can be taken by the institutions!

Some suggestions from my side:

  • Could you elaborate on the reasons why you removed the columns? You can explain what type of columns you are interested in to answer the project’s questions
  • You can shorten and simplify the function diss() if you just apply it to separate columns and not to the entire dataset. Can you see the way to do it?
  • This way you will also make the function universal and appliable to both datasets. Do duplicate function names! It will only confuse the reader and yourself especially if you have to develop a more complicated project
  • You have a lot of code repetition in [25] and [26]. Could you find a way to avoid this?
  • In [28] note that you have either text or a hyphen, - as the values. Instead of repeating the same code, you can write a function
  • It’s not very clear to me why you are including Study, Travel, etc. Do you consider them as types of dissatisfaction?
  • %matplotlib inline - place this code in the first code cell
  • Don’t forget about titles and axes labels for your plots. They are critical for understanding
  • Also consider using horizontal bar plots when tick labels (like ages in [34]) are too long
  • Also use colors wisely. In the plots description, you say that we can easily observe that for Dete resignations, 56 or older is most dissatisfied and for Tafe resignations, 41-45,46-50 and 51-55 groups are most dissatisfied.. It’s not very easy to observe at the first glance. You can make these bars colourful while the other bars gray. Also consider adding a subtitle directly in the plot where you write your comments
  • You code style is not always consistent. If you are using Jupyter Lab, you can install Jupyter Lab Code Formatter (GitHub) which custom code formatters (like black) will automatically and nicely format the code
  • In [38] rotate the tick labels by 90 degrees to make them more readable
  • The pie charts Dete Resignations by Employment Statusand Dete/Tafe Resignations by Year not readable because of overlapping values. Could you come up with a solution?
  • I would also make years integers and not floats because it looks more natural
  • You say Much more females resigned than males overall, due to some kind of dissatisfaction - so what step should be taken by the institutes?
  • if their salaries are competent - I guess you meant competititve:slight_smile:

I hope this feedback is helpful and look forward to seeing what the other learner has to add.

Happy coding :grinning:

1 Like

Thank you so much @artur.sannikov96 for this feedback. My project definitely needed some improvements. Could you please elaborate on the code style or share any resource that explains the same?

This website could be a good start. It contains a lot of information on how to write the code. If you want some guides on function documentation, pandas GitHub is a great source. They have high-quality documentation (in docstrings) for each of their functions. For example, have a look at the function to_numpy() here and then in their official documentation. I tend to use their style for the documentation in my projects:)

P.S. Don’t just blindly follow PEP-8 because it may be misleading. Have a look at this talk by Raymond Hettinger.

I hope these resources are useful to you!