Water Quality Dataset from Water Sources in Kenya

This dataset contains water quality measurements collected using an IoT-based sensing system deployed across multiple water sources in Kenya. The data was gathered over February–March 2026 at two-minute intervals, resulting in 100,800 observations.

Each record includes four key physicochemical parameters: pH, Total Dissolved Solids (TDS), turbidity, and temperature, a binary classification label indicating whether the water is safe for drinking or requires treatment.

The dataset covers diverse water sources, including:

Treated municipal and bottled water (safe)
Rivers and lakes (surface water)
Borehole water (groundwater)
Marine water (Indian Ocean)

Preprocessing steps included handling missing values through imputation, outlier assessment based on environmental plausibility, and feature normalization.

The dataset is designed for machine learning applications, particularly binary classification of water quality and real-time monitoring using IoT or edge (TinyML) systems.