Test Plan Short Term Dev Plan ============================= DHD20160709 1) Write front-end that takes Navetas-style bulk energy data and HDD data (and optional date range) for DHD (ie publishable data set) and reproduce test vector results. Inputs should be: a) bulk HDD data covering desired range b) bulk kWh/day data covering all households of interest c) optional date range d) optional subset of households to process Output should be simple kWh/HDD and other variables per household. ***Possibly consider change in R^2 between daily and weekly sampling to get an idea of regularity of weekly routines for other heating fuel uses. [DHD20160806] 2) Consider producing synthetic clean and noisy test data to verify calculations. 3) Write routine works out when energy-saving features are enabled given: a) OpenTRV valve (and boiler control) JSON logs a) groupings of IDs per household 4) Compute OpenTRV efficacy with inputs: a) cross-reference file between OpenTRV and Navetas households b) (1) and (3) 5) Review calculations, tests, methods, etc, and consider methods to discard outliers and signal features such as low R^2. DHD20160714: 1) ETVPerHouseholdComputation needs (at least two) implementations: 1a) Simple (no efficacy measurement) overall kWh/HDD estimate. 1b) Full (log parsing, efficacy measuring) + confidence over household set 2) Output generator (from ETVPerHouseholdComputationResult / set of): 2a) Plain text report 2b) Machine-friendly for further processing (eg to do/allow confidence comps) 3) Driver to scoop up all data + cross-referencing and produce report. Somewhere should be control to manually exclude some outliers or results influenced by external events such as change of tenants or secondary heating or boiler failure, etc, and parameters to adjust automatic filtering.