Live Fish Transportation trip data

About Dataset

Here's the "About" section for the live fish trip dataset upload: About (Dataset in Brief) Overview This dataset captures real-world, time-series water-quality and journey data collected during live fish transportation trips, monitored using our proprietary BlueSense IoT units. It was generated as part of the NEEL VAHAN AI-IoT platform deployment for live fish transport, developed under the IndiaAI Innovation Challenge (ICV1). What the Dataset Contains Each record in the dataset corresponds to a transport journey and includes continuous time-series sensor readings captured throughout the trip: pH, dissolved oxygen (DO), temperature, and total dissolved solids (TDS), logged at regular intervals from origin to destination. Alongside the sensor stream, each trip is tagged with journey metadata — route, duration, stocking density, and vehicle/tank details — as well as trip-level outcomes, primarily fish mortality rate, which serves as the key label for downstream predictive modeling. How It Was Collected Data was collected from BlueSense units deployed across multiple farms and transport routes, covering a range of real operating conditions rather than a single controlled environment. This includes variation in water source quality, ambient temperature, trip duration, and stocking density, which together shape the stress conditions fish experience in transit. As of the current deployment stage, the dataset spans multiple farms/routes and a growing number of monitored journeys, with continuous ingestion as new trips are completed. Intended Use The dataset is designed to support two connected modeling goals. First, it underpins our AI-driven soft-sensing models for NH3 and CO2, providing the paired pH/DO/temperature/TDS time-series needed to train and validate models that infer these otherwise-expensive-to-measure parameters from existing sensor data. Second, it supports predictive mortality modeling — correlating water-quality trends and journey conditions with actual mortality outcomes to enable early-warning systems that flag high-risk trips before losses occur. Why It's Valuable Paired, real-world water-quality and outcome data from live fish transport is scarce, since most transporters lack any continuous monitoring infrastructure at all. This dataset is unusual in combining continuous multi-parameter sensor telemetry with trip-level ground-truth outcomes (mortality), collected under our own deployed hardware rather than simulated conditions. This makes it directly usable for training soft-sensing and predictive models that generalize to real operating environments, rather than idealized lab conditions — supporting the broader goal of making predictive, data-driven fish transport management accessible to small and mid-scale operators.