This project aims to analyze and understand the global gap in mathematics proficiency among primary school students before and after the COVID-19 pandemic. The work focuses on comparing learning outcomes between 2019 and 2023 and exploring the possible role of factors such as:
- School closures
- Education spending
- Digital connectivity
- Access to trained teachers
Our team brings diverse perspectives from teaching, data analysis, and lived experience:
- How much did math proficiency change during the pandemic?
- Which countries or regions were most affected?
- What factors are most strongly associated with greater learning loss?
for more information about background research and problem framing question you can find it in this (folder) in this (file)
📌 Problem Statement
During our group discussion, we shared personal experiences of how COVID-19 disrupted math learning in different but connected ways:
- 👩🏫 Teachers (e.g., Heba) struggled with remote classes — unstable internet, shared devices, and large classrooms later made it impossible to support every student.
- 👩🎓 Students (e.g., Nada) faced poor connections, unclear recorded lessons, and lack of support, leaving them behind in math and other subjects.
- 🌍 Context (e.g., May from Myanmar) showed how the pandemic, combined with political crises, forced students to delay or stop their education.
- 🤝 Peers (e.g., Jubayer & Alexander) highlighted similar struggles in access and quality across different regions.
💡 Together, these stories show serious and unequal math learning gaps, especially in low and middle-income settings. Teachers were overwhelmed, and students lacked even the most basic learning tools.
you can find more information in this (file)
- Quantify the change in math proficiency across countries.
- Identify patterns and factors that may have contributed to the learning gap.
- Provide insights through visualizations and models to better understand the educational impact of the pandemic.
Diagram showing the data flow and analysis process to explore the math learning gap post-COVID.
Our data was mainly collected from UNESCO’s UIS (Institute for Statistics) open data platform.
To add important context, we included supplementary datasets from UNICEF (digital connectivity)
and the World Bank (income classifications and education indicators).
-
🎯 Focus: % of students achieving minimum math proficiency
-
🌍 Grouping: Countries by World Bank income levels
-
📉 Case study: Bangladesh school data (2019–2021) confirmed visible learning loss
👉 For full details on data sources, variables, and limitations (folder)
We cleaned and standardized the raw datasets to make them ready for analysis.
- 🗂️ Converted raw data into a consistent format
- 🚫 Removed non-country entries
- 🌍 Aligned all records using ISO country codes
- 🔗 Merged variables into a single dataset
👉 The cleaned output is available as:(file) 1_datasets/final_dataset.csv
We explored both the global dataset and a case study dataset to understand learning patterns.
-
🌐 Global dataset (
data_exploration.ipynb):
Checked structure, missing values, descriptive stats, and visualized math proficiency changes (2019–2023).
Also identified countries with the largest learning losses. -
🇧🇩 Bangladesh case study (
class7_math_exploration.ipynb):
Explored grade 7 math scores (2019–2021) with boxplots and histograms to observe the pandemic’s disruption.
👉 Full details and visuals are in the exploration notebooks. here class 7 and (here global)
We analyzed the dataset (2019–2023) to understand how math proficiency changed during the COVID-19 period and which factors explain this variation.
- 🔍 Explored correlations between key education indicators
- 📈 Built linear & multiple regression models
- 🖼️ Visualized results using heatmaps, regression plots, and residual checks
👉 For full details, methods, and code, see Analysis Notebooks.
- Over 1.6 billion learners worldwide experienced educational disruption, with 463 million children unable to access remote learning.
- Field studies documented sharp declines in mathematics proficiency, including a >27% drop for Brazilian second graders and regression to pre‑primary skill levels in Kenya and Uganda.
- The affected cohort may face up to $17 trillion in lost lifetime earnings, and learning poverty in low- and middle-income countries could reach 70%.
- Structural issues like weak digital infrastructure, teacher burnout, and inflexible curricula deepened the gap.
- Country data show divergent paths between 2019 and 2023: Albania and Armenia improved, while Australia, Azerbaijan, and Belgium declined.
A policy-focused plan targeting education ministers, advisors, and international organizations.
Core message: Targeted math recovery programs can close 78% of COVID-19 gaps at 1/3 the cost of grade repetition (Bangladesh + UNESCO data).
Deliverables:
This work is part of an academic data science initiative focused on understanding global development challenges using real-world data.