Skip to main content

Table 2 Basic differences between Autoflow© and AutoflowU

From: An adaptive model for the autonomous monitoring and management of water end use

Features Autoflow© Autoflow U
Version 2.1 Version 3.1 Version 1.0
Released date December 2013 September 2018 November 2018
Model development Rely on available water end use data for model training Do not require any existing data for development and calibration
Applied techniques • Hidden Markov Model
• Dynamic Time Warping Algorithm
• Time of day Probability
• Hidden Markov Model
• Dynamic Time Warping Algorithm
• Deep Learning (Stacked Auto-encoder)
• Time of day Probability
• K-means clustering
• Self Organising Map
• Dynamic Time Warping algorithm
• Self Organising Map
• Decision Tree Method
Available categories for categorisation 8 – Shower, Bath, Irrigation, Clothes washer, Dishwasher, Evaporative Cooler, Toilet, Tap   7 – Shower, Irrigation, Clothes washer, Dishwasher, Evaporative Cooler, Toilet, Tap.
(*) Shower and Bath are merged into one single category
Obtained accuracy when analysing different dataset
Australia data 75.0–92.0 76.3–93.5 72.1–92.3
US data 72.5–90.6 72.5–90.6 74.2–92.2