ICFAR Agricultural Sensor Analytics Platform

The ICFAR Agricultural Sensor Analytics Platform is a Python-based software tool developed to improve the processing, validation, and interpretation of agricultural sensor data. The project was created to transform raw sensor logs into clearer, more usable information through structured tables, visualizations, and summary statistics.

During my summer work at ICFAR, I supported plant maintenance, greenhouse monitoring, and daily data collection connected to ongoing research. My responsibilities included watering plants, checking environmental and system conditions, reporting operational issues, and collecting measurements such as water usage, pH, and nutrient levels for reporting to the PhD researcher. Through this experience, I became interested in how software could reduce manual review and make research data easier to interpret and use.

To help address this need, I proposed and began developing a dashboard that could visualize sensor and environmental data more clearly by showing trends, temperatures, and summary statistics in an accessible format.

The platform was built in Python and processes raw sensor log files into a cleaner structured dataset. The data-processing pipeline removes invalid entries, skips malformed rows, handles noisy formatting, and loads the cleaned results into tabular form for analysis. The schema includes timestamps, eight temperature probes, four soil-moisture channels, four Teros millivolt readings, and four Pascal values.

The dashboard was built in Streamlit and supports interactive exploration of recent sensor activity through time-window filtering, temperature-probe selection, statistics views, tabular inspection, and CSV export. It includes line graphs for soil moisture and temperature, summary statistics for numeric fields, a box-plot view for selected metrics, a recent-overview table, and a full filtered table with optional advanced columns such as Pascal, Teros, and extra temperature probes.

• Raw sensor log cleaning and restructuring into tabular data
• Timestamp parsing and filtering over selectable time windows
• Soil-moisture trend graphs for VWC1–VWC4
• Temperature trend graph for selectable probes T01–T08
• Summary statistics including mean, median, min, max, and standard deviation
• Box-plot analysis for selected numeric metrics
• Overview and full-table views for filtered data
• Customizable advanced columns including Teros and Pascal readings
• Downloadable filtered CSV output

This project reflects my interest in using software, data processing, and visualization to support sustainability, agriculture, and environmental research. It was designed to improve how technical information can be processed, reviewed, and communicated in research-support settings, while reducing the effort required to inspect messy real-world sensor logs.

• Python
• Pandas
• Regex-based parsing
• Streamlit
• Plotly

• GitHub Repository
• Live Dashboard Demo