Statistical analysis and forecasting of Zimbabwe's inflation rate (2010β2025) using Python, Matplotlib and Power BI
# πΏπΌ Zimbabwe Inflation Analysis & ARIMA Forecasting (2010β2025)
> **Author:** Shareef Chitesi
> **Degree:** BSc Honours in Applied Mathematics and Computational Science
> **Institution:** Midlands State University β Year 2
> **Tools:** Python Β· Pandas Β· NumPy Β· Matplotlib Β· Seaborn Β· Power BI Β· Excel
---
## The Story Behind This Project
Zimbabwe has one of the most extraordinary inflation histories in the world. I grew up watching prices change overnight, seeing currencies come and go, and hearing adults talk about the economy in ways that never quite made sense to me as a child. When I started studying Applied Mathematics and Statistics at university, I finally had the tools to go back and actually *understand* what happened β and more importantly, to model where things might be heading.
This project is my attempt to do exactly that. Using real data from the World Bank, RBZ, and ZIMSTAT, I analysed 15 years of Zimbabwe's inflation history across three very different currency eras. I then built a full **ARIMA model from scratch** β without using any forecasting libraries β implementing the mathematics directly using NumPy's linear algebra tools. This means every AR coefficient, every AIC calculation, and every residual diagnostic was hand-coded using the same mathematical principles covered in my degree.
This isn't just an academic exercise. For banks, insurers, and financial institutions operating in Zimbabwe, understanding inflation behaviour is critical β for pricing products, managing risk, and planning for the future. This project is my way of showing that I can think about those problems quantitatively and rigorously.
---
## π What I Found
| Metric | Value |
|--------|-------|
| Period covered | 2010 β 2025 |
| Peak inflation | **557.2%** (2020 β ZWL hyperinflation) |
| Lowest point | **-2.4%** (2015 β deflation) |
| Mean inflation (full period) | **75.0%** |
| Median inflation | **7.2%** |
| Selected model | **ARIMA(4, 1, 0)** β chosen by AIC |
| AI β¦