Personal data analysis project on the commercial and critical performance of 34 Marvel Cinematic Universe (MCU) films released between 2008 and 2024 (Phases 1 to 5, from Iron Man to Deadpool & Wolverine).
The analysis is framed by a fictional business brief written for practice. It is not client work.
| Step | Tool | File |
|---|---|---|
| Profiling and data-quality checks | Python / Pandas | 01_profilage.py |
Cleaning (release_date converted to a date type) |
Python / Pandas | 02_nettoyage.py |
| Exploratory data analysis and charts | Python / Pandas / Matplotlib | 03_eda.py |
| Key indicators and insights | Python / Pandas | 04_insights.py |
| Star-schema model, DAX measures and dashboard | Power BI Desktop | PowerBI_Source/MCU_Data_Analytics.pbix |
| Analysis report (French and English) | LaTeX | PowerBI_Source/utbmciad-report/ |
The Python scripts and the reports are written in French; an English version of the report is provided.
- 34 films, 20 columns: release dates, MCU phase, director, lead actor, runtime, production budget, worldwide and domestic box office, profit, ROI, IMDb, Rotten Tomatoes and Metacritic scores.
- Public dataset downloaded from Kaggle. Its bundled documentation (
DATASET_README.md,data_dictionary.csv) lists IMDb, Box Office Mojo, Rotten Tomatoes and Wikipedia as sources. The values are used as provided and were not re-checked against those sources. - Files:
marvel_movies_dataset.csv(raw data used by the scripts), plus the same data in.jsonand.xlsx;PowerBI_Source/marvel_movies_dataset_clean.csv(cleaned output of02_nettoyage.py, used by Power BI). - Data-quality checks in
01_profilage.py: no missing values, no duplicates, 34 unique IDs, consistent phase / date fields, and profit, ROI and domestic share consistent with their formulas.
Figures produced by 04_insights.py (amounts in millions of US dollars):
- Total production budget 6,947; total worldwide box office 31,129; total calculated profit 24,182.
- Average ROI per film 339.58 %; overall ROI (total profit / total budget) 348.09 %.
- Phase 3 has the highest average worldwide box office (1,227.27), the highest overall ROI (481.65 %) and the highest average critic scores. Phase 5 has the lowest overall ROI (186.50 %).
- Correlation with worldwide box office: production budget 0.63, IMDb rating 0.66, Rotten Tomatoes 0.44, Metacritic 0.42.
These are descriptive associations on 34 films, not causal effects. "Profit" here is worldwide box office minus production budget: it excludes marketing, distribution and other costs, so it is not studio accounting profit.
Requirements: Python 3 and the packages in requirements.txt.
git clone https://github.com/MediJr/MCU-Data-Analytics.git
cd MCU-Data-Analytics
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txtRun the scripts from the repository root, in this order:
python 01_profilage.py
python 02_nettoyage.py
python 03_eda.py
python 04_insights.py02_nettoyage.pywritesmarvel_movies_dataset_clean.csvat the repository root;03_eda.pyand04_insights.pyread that file, so run step 2 first.03_eda.pyopens seven Matplotlib windows one after another; close each window to continue.- On Windows, if you redirect the output to a file, run the scripts with
python -X utf8to avoid encoding errors on accented characters and arrows. analysis_script.pyis the example script shipped with the dataset. It is not part of this project's pipeline and also needsseaborn.
Open PowerBI_Source/MCU_Data_Analytics.pbix in Power BI Desktop. The model is a star schema (Fact_Movie, Dim_Movie, Dim_Phase, Dim_Date) with DAX measures such as total box office, total profit, overall ROI and average critic scores. If Power BI cannot find the source file, point the data source to PowerBI_Source/marvel_movies_dataset_clean.csv.
Dashboard overview (all 34 films, no filter):
- French:
MCU_Analyse_Performances_Commerciales_Critiques_FR.pdf - English:
MCU_Commercial_Critical_Performance_Analysis_EN.pdf
Build instructions are in PowerBI_Source/utbmciad-report/README.md.
Lumu Bamana Manassé — github.com/MediJr
