The (V, I) characteristic in the .FLAC files is uncalibrated. Calibration files are provided in the top-level directory for the data set. The recommended procedure for processing the data is to convert from FLAC (a lossless audio compression) to WAV using an audio editing package such as sox or Matlab's audioread function. Next, convert from the [-1,1] range of values in the WAV file to volts and amps, using the calibration.cfg file for the house in question. This file specifies an amps_per_adc_step parameter and a volts_per_adc_step parameter; to calculate volts from the WAV files, use the formula: volts_per_adc_step * 2³¹ ADC steps * value_from_wav_file, and similarly for the current. (The data recorder uses 32-bit integers to record the waveforms as an audio signal; hence there are 2³² ADC steps for the full range from [-1,1] and 2³¹ ADC steps for half the range from [0,1] or [-1,0].) The 'phase_difference' parameter in the calibration file can safely be ignored since the measurement hardware introduces no significant phase shift.
Additional detailed metadata relating to the houses and their meters is available at http://data.ukedc.rl.ac.uk/simplebrowse/edc/efficiency/residential/EnergyConsumption/Domestic/UK-DALE-2015/UK-DALE-disaggregated/metadata

House 1 appliance channel numbers:
1 aggregate
2 boiler
3 solar_thermal_pump
4 laptop
5 washing_machine
6 dishwasher
7 tv
8 kitchen_lights
9 htpc
10 kettle
11 toaster
12 fridge
13 microwave
14 lcd_office
15 hifi_office
16 breadmaker
17 amp_livingroom
18 adsl_router
19 livingroom_s_lamp
20 soldering_iron
21 gigE_&_USBhub
22 hoover
23 kitchen_dt_lamp
24 bedroom_ds_lamp
25 lighting_circuit
26 livingroom_s_lamp2
27 iPad_charger
28 subwoofer_livingroom
29 livingroom_lamp_tv
30 DAB_radio_livingroom
31 kitchen_lamp2
32 kitchen_phone&stereo
33 utilityrm_lamp
34 samsung_charger
35 bedroom_d_lamp
36 coffee_machine
37 kitchen_radio
38 bedroom_chargers
39 hair_dryer
40 straighteners
41 iron
42 gas_oven
43 data_logger_pc
44 childs_table_lamp
45 childs_ds_lamp
46 baby_monitor_tx
47 battery_charger
48 office_lamp1
49 office_lamp2
50 office_lamp3
51 office_pc
52 office_fan
53 LED_printer

House 2 appliance channel numbers:
1 aggregate
2 laptop
3 monitor
4 speakers
5 server
6 router
7 server_hdd
8 kettle
9 rice_cooker
10 running_machine
11 laptop2
12 washing_machine
13 dish_washer
14 fridge
15 microwave
16 toaster
17 playstation
18 modem
19 cooker
20 modem_router

House 3 appliance channel numbers:
1 aggregate
2 kettle
3 electric_heater
4 laptop
5 projector

House 4 appliance channel numbers:
1 aggregate
2 tv_dvd_digibox_lamp
3 kettle_radio
4 gas_boiler
5 freezer
6 washing_machine_microwave_breadmaker

House 5 appliance channel numbers:
1 aggregate
2 stereo_speakers_bedroom
3 desktop computer
4 hairdryer
5 primary_tv
6 24_inch_lcd_bedroom
7 treadmill
8 network_attached_storage
9 core2_server
10 24_inch_lcd
11 PS4
12 steam_iron
13 nespresso_pixie
14 atom_pc
15 toaster
16 home_theatre_amp
17 sky_hd_box
18 kettle
19 fridge_freezer
20 oven
21 electric_hob
22 dishwasher
23 microwave
24 washer_dryer
25 vacuum_cleaner

Additional detailed metadata (using the NILM Metadata Schema) relating to the houses and their meters is available at
http://data.ukedc.rl.ac.uk/simplebrowse/edc/efficiency/residential/EnergyConsumption/Domestic/UK-DALE-2015/UK-DALE-disaggregated/metadata

Individual appliance monitors have a push-button switch to allow users to turn the connected appliance on and off. The activity of this switch is recorded in a channel_<X>_button_press.dat file. If the switch has just been toggled on then a `1' is recorded. If the switch has just been toggled off then a `0' is recorded. The motivation behind logging switch press events is that these provide (imperfect) room occupancy information. Switch on events should be a perfectly clean recording (i.e. the only possible reason for an on-switch event appearing in the data is that the user pressed the switch). Unfortunately, off-switch events may include false positives. Occasionally IAMs turn off spontaneously (an event which is impossible to distinguish from a genuine button press). Also, if power is lost and returned to the IAM within 12 seconds then this will be logged as an off-switch event. If the power is off for more than 12 seconds then the system assumes that the IAM was deliberately unplugged and hence the system will switch the IAM to its previous power state when it reappears; this automatic switch event is not recorded