Logo Lanfrica

korie-cyber/nigeria-power-analysis

Domaine:

environment and energy
Créateur:
kor
Hôte:
Data-driven analysis of Nigeria's national power grid using 731 days of generation data (2023-2024). Includes preprocessing, statistical testing, trend analysis, and anomaly detection. # Nigeria's Power Crisis: A Data-Driven Analysis (2023–2024) A data science capstone project analyzing 731 days of Nigeria's national power grid data to uncover generation trends, grid instability, and systemic deterioration between January 2023 and December 2024. --- ## Key Findings - Mean daily generation declined **8.6%** year-over-year (4,019 MWh → 3,674 MWh) - Grid variability increased **41%**, indicating chronic instability - Frequency violations occurred on **48.9% of days** in 2024 - Statistical anomalies increased **5.5x** from 2023 to 2024 - The 5 worst generation days in the entire dataset all occurred in **Q4 2024** --- ## Repository Structure ``` ├── data/ │ ├── nigeria_power_genertion_raw/ # Original unmodified dataset │ └── nigeria_power_generation_clean/ # Preprocessed and cleaned dataset ├── notebooks/ │ ├── Emmanuel_Olafisoye_Data_Cleaning_Notebook_Karatu_2nd_Semester.ipynb # Data cleaning and preparation │ └── Emmanuel_Olafisoye_Analysis_Notebook_2nd_Semester_Karatu.ipynb # Main analysis and visualizations ├── report/ │ ├── Emmanuel_Olafisoye_Proposal_Document_Karatu_Second_Semester.pdf # Project proposal document │ └── Emmanuel_Olafisoye_Final_Report_Karatu_Second_Semester.pdf # Full written report └── README.md ``` --- ## Project Documents | Document | Description | |---|---| | Project Proposal | Research questions, objectives, and methodology plan | | Final Report | Complete analysis, findings, and recommendations | --- ## Methodology | Technique | Purpose | |---|---| | Two-sample t-test | Confirm statistical significance of year-over-year decline | | Cohen's d | Measure practical effect size | | Linear regression | Quantify monthly trend slopes per year | | Z-score detection | Identify anomalous generation days | | Rolling averages | Smooth noise and reveal underlying trends | | Confidence intervals | Bound the true mean generation per year | --- ## Statistical Su …