- Repository:
challenge-data-analysis - Type of Challenge:
Consolidation - Duration:
4 days - Deadline:
25/04/2025 13:30 - Team challenge : 4
- Be able to use
pandas. - Be able to use Data visualization libraries.(
matplotlib,seaborn,plotly,...). - Be able to clean a dataset for analysis.
- Be able to establish conclusions about a dataset.
- Be able to find and answer creative questions about data.
- Be able to think outside the box.
The real estate company "ImmoEliza" wants to establish itself as the biggest one in all of Belgium. To pursue this goal, it needs to create a machine learning model to predict prices on Belgium's sales. That way, they can pick out the properties that are the most valuable to them.
But for this, it needs to do a preliminary analysis to gather some information. Having no in-house data scientist, they are looking for talented people to do it for them.
Since your last encounter with them went great, they reached out to you to do this job. Everything is in your hands now!
Take the dataset previously scraped to do the analysis. (If you were in different groups, decide together which dataset you are going to use.)
We have decided to use the Kangaroo dataset. So it's time to do a cleaning on it. A cleaned dataset is a dataset that doesn't contain any duplicates, is blank spaces or error-free. The rest of the analysis can be discarded if you neglect this step!
- No duplicates
- No blank spaces (ex:
" I love python "=>"I love python") - No errors
- No empty values, delete them or impute them
Now that the data has been collected and cleaned, it is time for the analysis. How many variables and inputs do you have? And so on...
Use the tools such as matplotlib/seaborn/plotly!
Answer the following questions with a vizualization if appropriate:
- How many rows and columns?
- What is the correlation between the variables and the price? (Why might that be?)
- How are variables correlated to each other? (Why?)
- Which variables have the greatest influence on the price?
- Which variables have the least influence on the price?
- How many qualitative and quantitative variables are there? How would you transform these values into numerical values?
- Percentage of missing values per column?
After analyzing your data, it's finally time to interpret your results. You have to communicate your analysis using simple words and a table or graph, then use the results to decide on your best course of action.
Example of questions you should be able to answer to:
*NB: This is a non-exhaustive list. Try to make a maximum of interpretations of the dataset. Bonus points from yours truly for creative and outside the box questions that you answer.*
You must be able to answer :
- Plot the outliers.
- Which variables would you delete and why ?
- Represent the number of properties according to their surface using a histogram.
- In your opinion, which 5 variables are the most important and why?
- What are the most expensive municipalities in Belgium? (Average price, median price, price per square meter)
- What are the most expensive municipalities in Wallonia? (Average price, median price, price per square meter)
- What are the most expensive municipalities in Flanders? (Average price, median price, price per square meter)
- What are the less expensive municipalities in Belgium? (Average price, median price, price per square meter)
- What are the less expensive municipalities in Wallonia? (Average price, median price, price per square meter)
- What are the less expensive municipalities in Flanders? (Average price, median price, price per square meter)
As a deliverable, we expected a compelling presentation that tells a story from the data. It could be in notebook format or any form of presentation.
Your code should be clean, structured and documented as always.
- Pimp up the README file:
- Description
- Installation
- Usage
- (Visuals)
- (Contributors)
- (Timeline)
- (Personal situation)
- Have a clean presentation
- Have a clean dataset
- Bonus, Have a look at the Pipeline object in scikit learn and try to replicate your Step 1 transformations cleanly.
- Create the repository
- Study the request (What & Why ?)
- Identify technical challenges (How ?)
- Title
- Legend
- Axis labels (do not forget units)
- Correct usage of colors
- Comparable scales
- No overlapping text
- No screenshots
| Criteria | Indicator | Yes/No |
|---|---|---|
| 1. Is complete | Know how to answer all the above questions. | |
Use pandas and matplotlib/seaborn |
||
| Nice presentation on the subject | ||
| Code is clean and structured | ||
| README is complete and nicely formatted | ||
| 2. Is great | Additional questions were answered. | |
| Bonus was answered. | ||
| The colors are chosen carefully. |
“The lottery is a tax on people who don't understand the statistics.” - Anonymous
