The aim of this project is to identify countries with a strong potential for online e-learning for adults with a high level of education (at least a baccalaureat), through :
- selection of suitable indicators from the world bank (explore at https://datatopics.worldbank.org/education/);
- exploratory data analysis of these indicators
This is project 2 for the Master in Data Science (in French, BAC+5) from OpenClassrooms.
The project demonstrates the inital steps in a data science project :
- configuring a python environment, and using a jupyter notebook
- importing, joining and describing the files in a dataset
- validating, cleaning and selecting relevant data
- elementary exploratory data analysis (distributions, correlations, temporal evolutions)
The notebook includes a list of its own requirements, and a procedure for pip install
of any missing libraries. It also contains procedures to download and unzip the data, if not already downloaded.
Data : The dataset of education indicators used in this analysis can be downloaded (37Mb) from the worldbank at https://databank.worldbank.org/data/download/Edstats_csv.zip.
Python libraries : numpy, pandas, matplotlib, seaborn, missingno, plotly, tqdm
- P2_01_notebook.ipynb : import, cleaning, exploratory data analysis
- P2_02_support.pdf : a presentation of the project
Note : Files are in French. As requested by the jury, the notebook has not been cleaned up : the focus is on data manipulation and exploration
The net income for a company offering e-learning is assumed to depend on :
- The population selected for marketing (a country, a town, a field of study)
- The percentage of this population with access to e-learning
- The percentage of this population educated to at least baccalauréat
- The driving factors for e-learning (for example: unemployment, rurality, gender parity, availabilty of courses)
- The net price the students are willing and able to pay for e-learning
An indicator is selected for each of these factors.
The evolution of these indicators for different e-learning business models are explored :
- with / without expensive student accompanyment
- with / without reconversion of the workforce (academic vs professional courses)
Based on the selected indicators, the countries with the highest potential are found to depend strongly on the business model
- python, data-preprocessing, exploratory data analysis, data visualisation
- correlation heatmap, pairplot, barplot, lineplot, jointplot, chloropleth
- business model, performance indicator
- Set up a Python environment
- Use a Jupyter notebook to facilitate code writing and collaboration
- Manipulate data with specialized Python libraries
- Master the fundamental operations of the Python language for Data Science
- Perform a graphical representation using a suitable Python library