March 2026 — Current
Version 0.3 — ML Engine
Launch v0.3 Dashboard →Complete engine rebuild powered by Random Forest feature importance. Data extracted from 90+ peer-reviewed papers, 51 real cleaning technologies ranked via 0-1 normalized 5-variable FoM with median imputation.
\[ FoM_{v0.3} = \frac{(AI\_Level)^{\alpha} \times (Automation)^{\beta} \times (Lifetime)^{\gamma}}{(OPEX)^{\delta} \times (Scratch)^{\epsilon}} \]
- Random Forest selected Top 5 features from 25-column dataset
- Median imputation for missing values — all rows ranked
- Strict 0-1 normalization (÷ column max) eliminates scale bias
- Review/Survey papers filtered out — only physical technologies
- Interactive Chart.js bar chart with gold/silver/bronze ranking
February 2026
Version 0.2 (Legacy)
Launch v0.2 Interactive Dashboard →Shifted to a robust, strict multiplicative formula architecture based on Dr. Sameh's system engineering feedback. Includes parametric exponents allowing users to dynamically adjust output priorities.
\[ FoM_{v0.2} = \frac{(Recovery)^{\alpha} \times (Lifetime)^{\beta}}{(Water)^{\gamma} \times (CAPEX)^{\delta}} \]
- Renamed core tracker from "Efficiency" to "Recovery Percentage"
- Introduced Lifetime and CAPEX parameters for commercial viability
- Implemented 1-100 feature scaling mapping to handle math variations safely
December 2025
Version 0.1
The original prototype developed during Semester 1. Focused heavily on additive logic to test and parse extracted literature components.
\[ FoM_{v0.1} = \eta + S \]
- Where \(\eta\) represented basic energy restored
- Where \(S\) represented a broad sustainability additive score
- Deprecated due to mathematical weight imbalance (1 Unit of Water != 1% Efficiency)