% METER data % Data Documentation % April 2019
Measuring and Evaluating Time-use and Electricity-use Relationships (METER) is a project collecting and depositing data on household electricity use, activities and other contextual and demographic information. This documentation explains the data collection process, the structure of the data and how to use it.
METER is a 5 year research project at the University of Oxford's Environmental Change Institute. It is funded by the Engineering and Physical Sciences Research Council (EPSRC) from 2015-2020 under grant number EP/M024652/1.
METER seeks to improve our understanding of household electricity use, its timing, flexibility and societal importance. By combining high resolution household electricity readings with simultaneous activity records of household members, METER data provides a platform for new analytical insights.
These include among others a more detailed understanding of socio-technical drivers behind
- high and low electricity consumption
- high peak time consumption
- ability to respond the load shifting interventions
- impact of interventions on individuals
- trends in electricity use over time and in response to adoption of new technologies and social practices
The data deposited in March 2019 is summarised in Table 1.
Table: Statistics of the METER data deposition March 2019
|Earliest record||17 Feb 2016|
|Latest record||18 Jan 2019|
|Complete electricity records||264|
|Electricity readings (1 minute)||596396|
|Electricity readings (10 minutes)||60085|
|Activity recorders issued||743|
|App based diaries||395|
A complete electricity record in Table 1 refers to records with good electricity readings. Some households attached the recorder shortly after the study period begins and still returned some valid data. Some of the registered households have not yet taken part, while others have done the study more than once.
This document has two main parts. Under Data collection the recruitment process and data collection instruments are explained. The Data deposition section introduces the data tables, their relationships and the meaning of fields, columns and values.
The data collection process involves the following stages:
Each of these steps is briefly explained here.
Thus far METER's participation has been on a voluntary opt-in basis. The following approaches have been used to promote the study:
- Local radio and television appearances
- Public engagement and visiting community events
- Posters (see poster.pdf) and flyers
- Working with project partners
- University press release
- Social media
- Snowballing (email existing participants to invite others, with the promise that the person with the most references wins a year free electricity)
- Collaboration with other studies and initiatives
From January 2018 onwards the Household survey captures the source of referrals.
No explicit rewards are offered, but each year one participant can win the cash equivalent of their electricity bill for the past year. See terms and conditions.
Consent for participation and data usage is sought at three points:
- At the end of the Household Survey
- Via email, prior to the parcel being sent out
- At the start of the Individual Survey
At the end of the Household Survey participants are made aware of the data policy as shown in Figure 1.
The data policy is available online and states:
You own the data
Electricity load data may seem innocuous, but it can contain a lot of personal information, such as when and for how long you were at home. In combination with the activity information even more personal insights are possible. We therefore treat these data as personal and sensitive.
It is important to us that you remain in charge of the data. If for any reason you become uncomfortable about us holding information from you, you are entitled to ask us to remove the data - no questions asked.
You owning the data also means that we will not share data without your consent.
By taking part in this study you consent to your data being used for research purposes. For any additional use, we will ask your consent first.
How we use your data
There are different levels of access we would grant:
Your name and address: These will not be shared under any circumstances. We may identify your region based on your post code, but not beyond street level.
Your survey information: These will only be used for research and may be made accessible to other researchers. Wider access is only available in aggregated or fragmented form. That means your information is combined with others, such that your individual data cannot be recognised.
Your activity information: Researchers will have access to these data. We may also like to share examples of activity data with the public, provided you have given consent explicitly. Otherwise we will only share aggregated data (i.e. combined with enough others that make it impossible to identify individuals).
Your load profile: We treat your electricity data with the same level of sensitivity as your activity data.
How we store your data
All of the data we collect, is stored on a secure server within a relational SQL database. Your name and address are stored separately from your survey information and your electricity readings. The only thing that links these data is an ID number we generate for you.
Finally, before participants complete the individual survey on the activity recorder, they have to agree to their data to be used for research, as shown in Figure 2.
Prospective participants register their interest to participate by completing a Household survey on energy-use.org.
Data collected includes (see Household table for questions and range of answers):
- Contact information
- Household composition
- Number and ages of household members (Figure 3)
- Typical occupancy
- House type
- Size of house
- Appliances (Figure 4)
- Household income
- Preferred date to take part
The electricity recorder is scheduled to automatically switch on and start recording at 5pm on a given day of the week (Monday to Sunday). The devices are therefore charged, configured and sent six days prior to the study day, such that their next start-up is the recording day. Participating households receive the envelope 2-4 days prior to their chosen date.
The envelope contains:
- A note with instructions explaining the content of the parcel and next steps (Figure 5)
- An activity recorder (Figure 6) for each household member aged 8 or above
- A booklet with instruction for the activity recorder (see booklet.pdf)
- An electricity recorder (Figure 11)
- A prepaid return envelope
Before January 2017 participants received a paper diary instead of an activity recorder (see diary.pdf).
Activity and electricity recordings are taken over a 28 hour period starting at 5pm, thus capturing two of the typically most energy intensive periods between 5pm and 7pm. We refer to this 28 hour period as a 'study day' (even though it has 4 more hours than a day).
Activity recorders (aMeters) are issues as a pre-installed app on an Alcatel Pixi 4 smart phone device. All other phone features are disabled, except the power button, which can toggle the device on/off. The devices are thus configured to only provide two functions:
The interface is specifically designed to be failsafe and intuitive. No sliders, gestures or nested menus exist. All operations are a choice of 6 large buttons on a 2 by 3 grid (Figure 7). The only exception is the home screen where reported activities are chronologically listed in scrollable form.
The aMeter prompts each participant to complete an Individual Survey covering:
- Consent for data to be used for research
- Personal information
- Age range (Figure 7)
- Occupation type
- Working hours
- Flexibility of working hours
- Working from home
- Financial - Contribute to electricity bills - Personal income
- Frequency/duration of use - Appliances (Dishwasher, washing machine...) - Activities (Screen time, cooking, exercise...)
See detailed questions and data structure in Section Individual.
Once the survey is completed the first star lights up on the activity recorder and the reporting of activities will lead to up to 5 stars being earned for 25 activities.
From the home screen a new activity can be entered via the "now" button. If the event is in the past or future, the "recent" button allows for the time to be adjusted, before the entry sequence begins.
The entry sequence follows the standard pattern in Figure 8.
An example of a possible entry sequence is shown in Figure 9.
The decision tree with six branches per screen quickly leads to a detailed description of activities.
Editing an existing activity
Tapping an activity on the home screen opens up the 6 editing options shown in Figure 10. These can be:
- add another activity that took place in parallel
- repeat that (past) activity now
- change the caption of the activity (time use code remains unchanged)
- change the time of the activity
- report the activity as ended
- delete the activity
Section Activities explains how activity records are processed and stored.
The app keeps function and content separated. All button labels, captions and help files are stored as plain text json files which can be modified or translated without coding experience. A German version of the app has been developed and further internationalisation and customisation is encouraged. The files containing the sequence of screens and buttons with labels, icons and time-use codes is also in the public domain (github:MeterActivities).
Electricity recorders (eMeters) are preconfigured to record electricity over a 28 hour period. Participants attach the current clamp beneath their mains electricity meter prior to the study period. Instructions are printed on the case. The eMeter is fully automated and has no switches. It does not provide feedback.
Once the devices are returned in a pre-paid envelope the data is downloaded. Units are then recharged and re-deployed with the next participant. For each user the eMeter is issued with a new ID (see Meta Table).
The 50Hz current clamp signal is recorded at 44kHz and sampled down to RMS readings for every second. Values are stored as Watt readings in a csv text file on the device.
While temporal accuracy is high, the recordings are prone to variations between devices (~5% error) and distortions when the power factor of loads is low. Some devices, especially motors with poor power factor correction, can lead to artificially high readings. To correct for this error it would be necessary to also record the voltage. This would require users to make a physical connection (for example in the fuse box). This has been avoided thus far.
In their current form, the duration of electricity recording is limited to ~42 hours by the battery life of the devices. With booster batteries this duration could be extended.
The main failure mechanism for electricity recorders are:
- tearing of the flexure on the current clamp, often due to difficult access under the meter
- damage to the iron core surrounding the wire, when closing the clamp onto (rather than around) the wire
The electricity recorder is covered by a patent held by Oxford Innovations.
Within days after returning the equipment, households receive a link to an interactive visualisation of their electricity profile and activities.
The hour of highest electricity consumption is annotated with "Your peak demand". Participants have the opportunity to provide additional information about appliances that may have been in use during this hour from a drop down menu. Simple statistics about minimum, average and peak usage are shown and compared with the study averages.
This Section explains the data structure and how to read and link records, using the following conventions:
- Table: a collection of records (shown in bold)
Field: a column in a Table (
- Record: row or set of data identified by an ID (italic)
The naming convention for a primary key (PK) of Table X is
except for the Electricity tables, where the primary key is always
idElectricity, even for table Electricity_1min and Electricity_10min.
Fields referring to a primary key in another table Y (foreign keys, or FK) follow the naming convention
The data relationships between tables in the Meter database are shown in Figure 13.
idContact is generated when a contact registers and their personal information is stored in Contact.
The Household Survey information is linked to the contact via the foreign key
Contact_idContact in Household.
Each time an
idContact takes part in the study, a unique
idHousehold is generated, which has
idContact as its foreign key.
All survey information is copied to the new
idHousehold and participants can update content if their circumstances have changed.
Any personal information in Contact is excluded from this deposition. It is still possible to identify repeat participants via their common
idContact in the Household table.
Recordings in Activities, Individual and Electricity are linked to an
idHousehold in two steps:
#. Each record has a device
idMeta with which it was recorded. This is stored as a foreign key (
#. Meta links this ID with the Household via
Example: to get all activities for a given
idHousehold = X, one can query:
SELECT * FROM Activities JOIN Meta ON idMeta = Meta_idMeta WHERE Household_idHousehold = X
The following sections describe fields, the meaning of values, as well as other conventions for each table.
The Meta table keeps a record of devices issued. Each time a recorder (either activity or electricity) is sent out, it is assigned a unique ID (
Two types of devices are in use:
- Activity recorders: either paper diaries, or app based
- Electricity recorders: for mains electricity or PV circuits
The device type is identified by
DataType as shown in Table 2.
Table: Device types
SerialNumber conventions further distinguish the type of device as shown in Table 3
Table: Serial number conventions
|xx = month of study day,|
|yy = serial number,|
|d = default day to record (0=Sunday)|
|16 digit||Downloaded app|
|(serial number of mobile device)|
Quality value is intended as a means to exclude some data from analysis without deleting it from the database.
For activity recorders
Quality is automatically assigned to the number of activities reported. If entries are considered erroneous, this value can be set to 0 to exclude them from analysis. A threshold of
Quality > 6 is recommended, but there is no reason why records with fewer entries may not be valid for some types of analysis.
Electricity and PV recorders get their
Quality value automatically assigned from the number of hours for which every 10 minute average is 20 Watts or above. Subsequently, profiles are visually inspected for unreasonable readings, resulting from noise or incorrect attachment of the clamp, in which case
Quality is set to 0. A good record should have
Quality > 26 (hours).
CollectionDate refers to the date of first reading for Electricity, or earliest activity record for Activities.
The Household table is populated as part of the online Household Survey. The identity of a household is given by the foreign key
Contact_idContact (the Contact table itself is not available for privacy reasons).
Households participating more than once have a unique
idHousehold for each study day. The survey information is duplicated and can be modified by the participant if circumstances have changed. It is possible to identify repeat household via their common
Contact_idContact (unless they re-registered afresh).
status field shows how far a household has progressed through the study process. The states are listed in Table 4.
Table: Status progression through the study
|2||no date selected|
|4||awaiting confirmation of participation|
|6||devices returned and processed|
|7||data shared with contact|
|8||Follow-up Surveys provided|
|31||offered new dates|
page_number is the last page a contact filled in as part of the Household Survey. When returning to the survey, participants can continue on this page.
date_choice is the date on which a participant chose to take part in the study. It is at 5pm on this day that data collection should commence. It is possible for activities to be recorded before and after this date.
study field distinguishes different phases of the study as shown in Table 5. While the process has been consistent throughout, it may be desirable to separate out participants who used a paper diary, rather than the app.
Table: Study phases identified in Household
|1||early trials, 24 hours|
|2||activities entered using paper diaries|
|3||activities entered using the METER app|
|4||community study (West Oxford)|
reference field was introduced in January 2018. Prior entries are NULL. The
referees field has been omitted from this deposition for confidentiality reasons. This field can contain the names of people who referred a participant to the study.
timestamp is the date and time when the record was created, i.e. when a participant first signed up.
The full Household table with all survey questions and answer ranges is shown in Table 6.
Table: Household table fields and values and original survey questions
|Field||Question and options||Values|
||How many people live in your household?|
|1, just me||1|
|6 or more||6|
|age_group||Your ages. How many of you in each age group?||#|
||Under 8 .. people|
||Age 8-18 ... people|
||Age 19-34 ... people|
||Age 35-50 ... people|
||Age 51-70 ... people|
||Over 70 ... people|
|Your pets. Click any pets you have in the house.||1: yes|
||Who's at home?|
|On a typical weekday how many people would be at home at 6pm?|
|5 or more||5|
||Your home. What type of house do you live in?|
|Other (please specify)||6|
||Rooms. How many rooms does your home have?|
|Not counting kitchen landing hallways and bathrooms(s)|
|6 or more||6|
||Your property. Do you own or rent your property?|
|Own with or without a mortgage||1|
|Rent social landlord||2|
|Rent private landlord||3|
|appliance_b||Your appliances. Which of these do you have?||1: yes|
|Click anything you have in the house.|
|Your appliances. How about these things? Click anything you have and use|
|appliance||Your appliances. Do you have any of these items?||#|
|If so click to say how many. Only count things you actually used in the last year.|
||Night storage heaters|
||Your electricity provider. Who provides your electricity?|
||Your tariff. Which of these best describes your electricity tariff? Just pick one|
||Your electricity bill. How much do you spend on electricity per year [month]|
|£300 [£25] or less||1|
|Up to £400 [£33]||2|
|Up to £550 [£46]||3|
|Up to £700 [£58]||4|
|Up to £850 [£71]||5|
|Up to £850 [£71]||6|
||Give bill as monthly figure||1|
||Income. What is your household income?|
|These are annual gross figures. Switch to monthly [annual]|
|[These are net monthly figures after deductions for tax and national insurance]|
|Less than £15,000  a year [month]||1|
|Less than £25,000  a year [month]||2|
|Less than £35,000  a year [month]||3|
|Less than £50,000  a year [month]||4|
|Less than £70,000  a year [month]||5|
|Over £70,000  a year [month]||6|
||Give income as monthly figure||1|
||How affordable do you find your electricity?|
|Select the statement which best describes the affordability of your electricity|
|Not affordable, I struggle to pay||1|
|Somewhat affordable, I must budge||2|
|Affordable,I can pay||3|
|Very affordable,I can easily pay||4|
||Which of these dates would suit you to take part?|
|none of these|
||How did you hear about this study?|
|Friend or other||5|
The default value is '0'. The default
date_choice is '2000-01-01' if no date has yet been selected. The
provider defaults to 'not given'.
p6pm, which could meaningfully be '0' has a default of '-1' to indicate a missing value. For some fields a missing value can be ambiguous. Appliances may be '0' because they are not present or because they were omitted when filling in the survey.
This table contains results from the Individual Survey. Any question can be skipped, in which case the default value is '0'.
Table: Individual survey questions and answer ranges
|Field||Question / options||value|
||What gender are you?|
||What age group are you?|
||What is your occupation?|
||How many hours do you work per week?|
||Are these hours always the same?|
|Always the same||1|
|Variable to suit work||2|
|Flexible to suit me||3|
||How often do you work from home?|
||Do you pay the electricity bills?|
|Yes, I do||1|
||Roughly, what is your annual income?|
|£0 - £10,000||1|
|up to £20,000||2|
|up to £30,000||3|
|up to £45,000||4|
||In the last 7 days, did you personally use a Dishwasher?|
|8 or more||6|
||In the last 7 days did you personally use a Washing Machine?|
|8 or more||6|
||In the last 7 days did you personally use a Tumble Dryer?|
|8 or more||6|
||In the last 7 days did you use a Shower?|
|8 or more||6|
||In the last 7 days did you use a Vacuum Cleaner?|
|8 or more||6|
||On a typical weekday how much time do you spend on Mobile Devices (in total)|
|10min or less||2|
|30min or less||3|
|1h or less||4|
|2h or less||5|
||On a typical weekday how much time do you spend on Screens (any device)|
|30min or less||2|
|1h or less||3|
|2h or less||4|
|3h or less||5|
||On a typical weekday how much time do you spend Cooking|
|30min or less||2|
|1h or less||3|
|2h or less||4|
|3h or less||5|
||On a typical weekday how much time do you spend Exercising|
|10min or less||2|
|10min or less||3|
|1h or less||4|
|3h or less||5|
||How interested are you in energy issues?|
|Not very much||4|
|Not at all||5|
All activities are precoded on the aMeter. Paper diaries were processed using the same app and coding structure.
Table: Activity table fields and value ranges
||Recording device ID||int|
||Time of activity||DATETIME|
||Time of reporting||DATETIME|
||Where are you?|
|- public transport||24|
||What are you doing?||text|
||Time Use Code||0..9999|
||How many people are you doing this with?|
|More than 4||5|
||How much are you enjoying this?|
|Not at all||1|
||Entry sequence||tuc ID list|
Default value is 0, unless this is an option, as for
people, where the default is '-1' instead.
Electricity consumption is recorded every second and downsampled to mean values every 1 minute and 10 minutes (Electricity_1min, Electricity_10min). The
Watt values correspond to the period beginning at
dt, i.e. they span from
dt + 10 minutes.
Table: Electricity table fields
||Recording device ID|
||Time of reading (datetime)|
||Power in Watt (float)|
A full record covers 28 hours, starting at 5 pm. Prior to 18 April 2016 records were only 24 hours, starting at 4 am.
The noise level of the recorder is 2-5 Watt. Readings of less than 20 Watt tend to suggest that the recorder was not attached properly.
The Legend table gives access to many of the survey questions and field meanings provided in this documentation. These can be convenient to labelling results during analysis. It describes field entries in the Household, Individual, and Activities tables.
tab contains the name of the table and
col the name of the field.
meaning pairs give the meaning for a range of possible values for this field. Where the range of values is open (counts and free text), no legend is provided.
The survey questions that were asked of participants is denoted with
Example: the meaning of a value '1' in Household field
SELECT meaning FROM Legend WHERE `tab` == 'Household' AND `col` == 'own' AND `value` == '1';
returns: "Own, With or without a mortgage"
Or, to establish the original question for
people, one can query:
SELECT meaning FROM Legend WHERE `tab` == 'Household' AND `col` == 'people' AND `value` == 'q';
returns: "How many people live at this address?"
Note that questions are truncated to 45 characters. For full wording of questions, refer to this documentation.
Table: Legend table with meaning of values in other tables
||Name of reference table: Household , Individual or Activities|
||Field in reference table|
||Range of values ('q' = survey question)|
||Explanation of corresponding entry|
The following publications give an introduction to the METER study.
Philipp Grunewald, Russell Layberry, ECEEE Summer Proceedings (2015). Measuring the relationship between time-use and electricity consumption
Introducing the methodology for recording activities and electricity alongside each other.
Philipp Grunewald, Marina Diakonova, Davide Zilli, Jessica Bernard, Adriano Matousek, ECEEE Summer Proceedings (2017). What we do matters – a time-use app to capture energy relevant activities
An introduction to the activity recording app and its features
Phil Grunewald, Marina Diakonova, Energy and Buildings (2018) The electricity footprint of household activities - implications for demand models
Outlines initial findings and tests the strength of activities to predict electricity use. We conclude that for multi-occupant households in particular the multiplicity of parallel activities does not support the use of activities as a strong indicator of household electricity use. The difference between activities with 'high' and 'low' electricity use is only around 7% at aggregated household level.
Philipp Grunewald, Marina Diakonova, Energy Research & Social Science (38) 58-66 (2018). Flexibility, dynamism and diversity in energy supply and demand: A critical review
The importance of everyday life limiting even largely technical solutions for load shifting is brought into focus through a systematic review of flexibility options.
Further publications can be found on Academia.edu: Academia
Meter project developments are disseminated online via:
This documentation is maintained on GitHub MeterData repository.