# Need some reviews on Guided project: Ebay car sales

Hi everyone,

It’s been a while I was able to post a guided project for review.

When Numpy and Pandas started, things became a bit more serious and needed some time to get the interest back and started finding the fun. So I think I had some fun with playing ebay used car data.

So after going through the instructions and solution notebook, I have decided to do some analysis by myself without exactly following the instructions.

I really had fun trying out new things and playing with the data. But I’m not sure how much of that fun actually translated into some good data analysis.

If anyone of you have some time to spare, please go through my project and let me know if it makes any sense, or is there a better and more efficient way to achieve similar results.

I’d also like to know how to combine/compare two different data frames with different index. For example, in this analysis how can I compare the price column with mean price of each brand. Price is a column in auto dataframe while mean price is a column in newly created brand_data dataframe. Index of auto is numbered while that of brand_data contains the name of each unique brands) (I hope some of you understood what I was trying to explain)

I am looking forward to learning from your feedback and suggestions.

Here is my ebay Used car sales Notebook
P3_Exploring Ebay Car Sales Data Ver 2.ipynb (199.4 KB)

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

3 Likes

Nice!
Loved your analysis it was interesting to read and loved the fact that you did things completely different. However i think that some plots are necessary, sometimes it’s easier to understand that way.

Regarding your question, i guess you can use the df.merge() method

Hope it helps you
Good luck!

3 Likes

Thank you @alegiraldo666 for the encouragement. This mean a lot.
And yes, I think plots can be a good idea. But I am yet to learn about that in the next course. I think same goes with df.merge() method also. Haven’t seen it in this course. So I think I can come back to this project once I learn them.

Thanks again for going through my project.

1 Like

Hi again @jithins123,

Finally I have looked through your project, and I can say that it’s really impressive and profound, congratulations! Very detailed analysis, interesting insights, creative approach to the project itself. This my feedback I will start with listing what I liked most

• Good idea to inspect the only different value in the columns like ‘seller’ before just dropping it. There could be a valuabe information at least to keep in mind and probably to get some insights. Well, it was not the case in given project, but you did well to check it anyway.
• Very good practice to check first all the extreme values and normal tendencies on Google. I especially liked that you googled the maximum price of a brand new limousine and in this way found some discrepancies in the dataframe. Also checking when certain brands were founded and comparing with our data was a cool idea.
• Using the method value_counts with bins - great, especially when there are a lot of unique values.
• I also liked that you analyzed power_ps values and its outliers, registration month, and calculated age and time taken to sell the cars.
• Very interesting part of analysis where you investigate cars that were difficult to sell and those easy to sell, the reasons for both and comparison with mean prices. Your assumption that some of the “easy-to-sell” listings were just deleted from the table for some reasons is reasonable and to the point.

Now I have some suggestions about what can be improved.

• The major issue is that something happened in the code cell [46], it was not calculated, and hence neither did the following code cells. Thanks goodness it happened close to the end of the project, but anyway you’d better re-run all the cells for not losing the last results. The insights afterwards look interesting, but the code results are not presented.
• At the beginning of the project it would be helpful to put a link to the original dataset.
• In case of this project, it’s better convert prices and odometer measurements into int, not float.
• Well, you already know my opinion about ‘0’ prices and that they can easily exist
• In some markdown cells (like in those above the code cells [3] and [5]) I would omit technical details and descriptions of the methods used. Also because the structure of such methods can be clearly traced from the code cells directly below them.
• After the code cell [5] - too many subheadings, or probably they should be of different hierarchy.

For the rest, great project! I really enjoyed reading it and learned some good inights and ideas to keep in mind also for my future projects.

Please let me know when your next project is ready for review, now I am your regular reader

2 Likes

Thank you so much for your time and this very detailed review! This is so great. Glad you liked some of the ideas that I’ve used in this project.

Yes, I saw the error in 46th cell. I was actually doing day 1, day 2 etc. Then dropped them all and changed to day9. But while doing I forgot to update the whole of copy-pasted code block. I didn’t see that before uploading, but I have corrected on my file. And till now I didn’t even realise that I have presented a code that hadn’t run full!! Since there were markdown cell after every code cell, I didn’t even see the missing output cells! Thank you for pointing that out! Now you know, I can be quite careless at times!!

This is the very reason why I missed adding the link to dataset. I usually do it.

I will remember the point about converting to int and searching more about the free cars on the internet.

I understand your point regarding omitting the technical details. But I am trying to write these notebook from the perspective of a DQ student who doesn’t have much experience, just like me. I’d definitely like a solution notebook with more explanations so that I can learn. It also helps me while reviewing other’s projects. If that is what you were referring, I’d like to keep that style. If it was something else, please let me know.

I understand. Somewhere on the way I lost the hierarchy. It also happened because I was doing it over the period of many days and I lost track of my own system for headings and sub headings. So they were not the perfect hierarchy. I’ll keep that in mind for my future projects.

Thanks a lot for these valuable comments. Really means a lot. Thank you.

2 Likes