How ademeecee/time-series-predictions works
Project type and purpose
• This is a time-series predictive modeling project implemented in Jupyter notebooks.
• Purpose: forecast hourly PM2.5 levels in Lagos using historical air-quality data.
File structure
• Top-level items:
– lagos_air_quality.csv (raw data source)
– Report_on_Time_Series_Linear_Regression.ipynb (final report notebook)
– Time_Series_Linear_Regression.ipynb (data prep + model pipeline notebook)
– README.md (documentation)
• No separate frontend/backend code or services—this is a data-science pipeline contained in notebooks.
README for additional insights
• Workflow stages are spelled out: data collection, EDA, feature engineering, baseline model, predictive modeling, evaluation, visualization, stakeholder reporting.
• Technologies implied: Python, pandas, scikit-learn, plotting libraries (e.g., matplotlib or plotly).
a. Main components
1. Data Ingestion (lagos_air_quality.csv)
2. Data Preprocessing & Cleaning (resampling, imputation)
3. Feature Engineering (lag features, rolling averages)
4. Baseline Model Module (mean-based predictor)
5. Predictive Model Module (Linear Regression training + inference)
6. Model Evaluation (MAE, R² calculations)
7. Visualization & Reporting (interactive/ static plots, final report notebook)
8. Future-data inputs (weather APIs or expanded datasets)
b. Relationships & interactions
• Data flows from Data Ingestion → Preprocessing → Feature Engineering → Model Modules → Evaluation → Visualizations.
• The report notebook reads outputs from all prior steps.
c. Architectural patterns / design principles
• Linear pipeline pattern: each notebook cell group acts as a modular stage.
• Separation of concerns: distinct tasks (cleaning vs modeling vs reporting) in separate notebook sections/files.
• Extensibility: future insertion of additional data sources or model types (ARIMA, LSTM).
d. Key technologies & libraries
• pandas (data manipulation)
• numpy (numeric operations)
• scikit-learn (linear regression, metrics)
• Jupyter environment (notebooks)
• plotting libs (matplotlib/plotly)