Analysis of Data Quality in Digital Smart Cities: The Cases of Nantes,

Hamburg and Helsinki

José L. Hernández

, Ana Quijano

, Rubén García

, Pierre Nouaille

, Lukas Risch

Mikko Virtanen

and Ignacio de Miguel

Energy Division, CARTIF Technology Centre, Parque Tecnológico Boecillo 205, Boecillo, Spain

Direction Territoriale Ouest, CEREMA, Rue René Viviani, Nantes, France

Senate Chancellary – Department of European Affairs, Free and Hanseatic City of Hamburg, Hamburg, Germany

VTT Technical Research Centre of Finland Ltd Espoo, P.O. Box 1000, Finland

Optical Communications Group, University of Valladolid, Valladolid, Spain

ignacio.miguel@tel.uva.es

Keywords: Data Quality, Digitalisation, Smart Cities, Data, Completeness, Correctness.

Abstract: The Smart Cities concept is supported by the use of Information and Communication Technologies (ICT),

which enables the digitalisation of the city assets. Then, cities are nowadays driven by data, with a clear

dependency on the data collection approaches. Decisions and criteria for urban transformation therefore rely

on data and Key Performance Indicators. However, one question remains and refers the reliability and

credibility of data that guide the decision-making processes. Many efforts are made in the definition of the

data quality methodologies, but not in analysing the real situation about data collection is smart cities. This

paper applies a methodology to quantitatively analyse the real quality of the data-sets in the cities of Nantes,

Hamburg and Helsinki. This work is under the umbrella of mySMARTLife project (GA #731297). The main

conclusion or lessons learnt is the need for more appropriate methods to increase data quality, instead of

defining new methodologies. Data quality requires improvements to make better informed decisions and

obtain more credible Key Performance Indicators.

1 INTRODUCTION

Nowadays, cities are responsible for more than 60%

of greenhouse gas emissions (Eurostat-a, 2022). To

tackle this issue, the European Commission has

established ambitious plans to reduce the emissions in

55% in contrast to current practices by 2030 (2030

Climate & Energy Framework, 2022), reaching

climate neutrality by 2050.

For than end, cities need a transformation towards

Smart Cities and must integrate multiple perspective

and verticals such as energy, mobility, nature,

economy or water management, among others. All of

them supported by the integration of the Information

https://orcid.org/0000-0002-7621-2937

https://orcid.org/0000-0002-0366-7375

https://orcid.org/0000-0002-0358-8001

https://orcid.org/0000-0001-7982-5117

https://orcid.org/0000-0002-1084-1159

and Communication Technologies (ICT) (Batty,

2012), approaching digital cities.

Digitalisation relies on data and new technologies

like IoT (Internet of Things) to be able to monitor the

city assets. Nevertheless, main challenges lie in the

quality of the data and, thus, the accuracy and

confidence when making decisions.

Efforts are put on the definition of urban

platforms, acting not only as repository of

information, but ingesting, transforming data, as well

as calculating indicators and exposing useful

information to make better informed decisions. The

mySMARTLife EU project (mySMARTLife, 2022),

with GA #731297, works in this direction. The

project, with more than 150 actions taking place in the

Hernández, J., Quijano, A., García, R., Nouaille, P., Risch, L., Virtanen, M. and de Miguel, I.

Analysis of Data Quality in Digital Smart Cities: The Cases of Nantes, Hamburg and Helsinki.

DOI: 10.5220/0011271900003269

In Proceedings of the 11th International Conference on Data Science, Technology and Applications (DATA 2022), pages 353-360

ISBN: 978-989-758-583-8; ISSN: 2184-285X

353

cities of Nantes, Hamburg and Helsinki, aims at

reducing the energy demand of buildings, promoting

e-mobility and creating urban data platforms

following an open specifications framework

(Hernández, 2020). Under the scope of the project, all

the actions must be monitored with real data to extract

conclusions and calculate impacts, making data

quality essential. In this context, this paper presents

quantitative results of real data quality in urban city

platforms (Nantes, Hamburg and Helsinki) through

the application of a methodology developed within

mySMARTLife project.

The paper is organised as follows. Section 2

provides a background about data quality and the

existing analysis techniques. Section 3 describes the

methodology applied in mySMARTLife for data

quality. Section 4 continues with a set of examples

about completeness and correctness of data in the

three cities of Nantes, Hamburg and Helsinki. Section

5 extracts a set of conclusions and future work.

2 BACKGROUND

As it has been already introduced, cities are currently

working on their transformation to become more

resilient and climate neutral. However, how could

anyone determine the level of smartness or carbon

neutrality? The answer to this question is the

application of evaluation frameworks that are driven

by KPIs (Quijano, 2022). The calculation of the

indicators relies on real data, which, due to occasional

gaps, out of range and other errors in the collection

and processing, does not provide useful insights

(Alanne, 2021).Data quality is then crucial, not only

in the assessment, but also in the creation of

intelligent data-driven services (Hassan, 2021).

Data quality indicators could be split into several

groups (Schmidt, 2021):

 Integrity, which refers to whether data comply

with structural and technical requirements or

not.

 Completeness, which focuses on the avoidance

of data gaps according to the frequency and

expected distribution of data.

 Correctness, represented by consistency and

accuracy, which refers to out of range

identification, in other words, error-free data.

 Timeliness, i.e., how up to data to data-sets are.

 Interpretability, which means the extent to

which data can explain the reality.

 Accessibility, which, in the case of smart cities,

is the data availability via open data portals or

APIs (Application Programming Interface).

 Interoperability, which is described as the

ability to access and process data from

heterogeneous data sources.

Despite the efforts, data quality is still the main

challenge in the digitalisation of cities, but, the

reliability is questionable due to the data issues (e.g.

communication or infrastructure problems) (Sin

Yong Teng, 2021). Moreover, traceability of the

errors is not easy (Hossein Motlagh, 2020), mainly

taking the big amounts of data being collected into

account. Additionally, there is no consensus about

governance of data (Ender, 2021) and data-sets are

managed in silos, limiting the accessibility (Abraham,

2019).

Having all these aspects in mind, a methodology

to quantify the aforementioned quality indicators has

been defined in the framework of mySMARTLife and

applied in the cities of the project.

3 METHODOLOGY FOR DATA

QUALITY ASSESSMENT

mySMARTLife project has developed urban data

platforms in three lighthouse cities: Nantes, Hamburg

and Helsinki with an open specifications’ framework

and interoperability mechanisms and surveillance

modules. (Hernández, 2020).

Within the methodology approached in the

project, a statistical definition has been followed to

determine the level of quality for data in the urban

platforms, which is qualified by completeness and

correctness indicators defined as follows:

 Completeness is calculated as equation 1

Completeness = nc / ne * 100 (1

)

where the number of collected samples (nc) is

the counter of total samples stored in the

databases and the number of samples to be

expected (ne) is calculated as equation 2.

ne= freq * iter * time (2

)

The term freq is the data collection frequency,

while iter is the number of iterations and time

corresponds with the data collection span. For

instance, considering a frequency of 1 minute

along 1 hour, the period factor would be 60

iterations, with a total of expected samples

equal to 60.

 Correctness is determined by the values within

the range to be expected, as depicted in

equation 3.

min

≤x≤ x

max

)

DATA 2022 - 11th International Conference on Data Science, Technology and Applications

354

Where x

max

= upper limit and and x

min

= lower

limit. Here, the meaning of the values to be

expected should be remarked. In contrast to the

sensor range, which also sets up maximum and

minimum values, the value to be expected is the

one that is a normal quantity, between

reasonable lower and upper limits (x

min

, x

max

respectively). For example, an indoor

temperature sensor could measure -5ºC, which

is inside the physical sensor measurement

range, but it is considered as abnormal value.

In the case of energy meters, usually, these

measure cumulative values, so that infinite is

the maximum value and cannot have negative

values.

As established in the project, data quality

reporting is executed every 6 months, obtaining an

overview of the quality indicators in each data quality

report which helps to identify how suitable are data

collected for the KPIs calculation. Within each report,

if a data-set is identified as non-compliant with the

completeness and correctness criteria, granularity of

data report is reduced (i.e. from the 6-months

aggregation to monthly values) to determine the

reasons of the deviations with respect to the quality

criteria.

According to the timeline of mySMARTLife, data

collection started in December 2019, when the

interventions in the three cities finished (although

some actions present delays in the implementation).

Six reports are expected to be delivered during

mySMARTLife project duration.

 1st report (Dec‘19-May‘20)

 2nd report (June‘20-Nov‘20)

 3rd report (Dec‘20-May’21)

 4th report (Jun'21-Nov'21)

 5th report (Dec’21-May’22)

 6th report (June’22-Sept’22)

Four of them are reported at the time of writing

this paper and while the first three reports have been

fully analysed, only initial insights of the fourth report

are included.

According to (Araújo, 2017), it should be

indicated that data completeness can be below 100%,

being 80% a reasonable threshold to consider data as

compliant with quality requirements. mySMARTLife

has slightly resized this number and a traffic light

analysis has been performed where values of

completeness lower than 75% are considered non-

valid, values with more than 90% are very high-

quality data, whereas between 75% and 90% are

considered with enough quality for further KPI

calculation.

4 CASES OF NANTES,

HAMBURG AND HELSINKI

As introduced, around 150 actions have been

implemented in the three lighthouse cities in the

mySMARTLife project: Nantes (France), Hamburg

(Germany) and Helsinki (Finland) under the pillars

energy, mobility and ICT. Table 1 summarises the

type of project interventions in which data quality

method is applied.

Table 1: Lighthouses interventions in energy and mobility

pillars.

Category Interventions

Buildings High-performance districts for

Retrofitted and New buildings

Energy

systems

Digital boilers, PV, Organic PV films,

Hybrid solar power, Batteries

City

infrastructure

District heating with RES, Hydrogen

injection in district heating, Solar

power plant, Wind farm with storage,

Waste heat, Smart heating islands,

Heat pumps, Smart lighting

e-Mobility

Electrification of public fleet (XXL

eBus, Autonomous bus, e-cars, e-

bikes), Car sharing

Charging

infrastructure

Solar road, Smart/fast/renewable

charging stations

4.1 Nantes

The first batch of interventions selected comprises the

hybrid solar power system plus the retrofitting of

individual houses. This intervention consists of two

data-sets: renewable production by the panels in

electricity (i.e. elec_prod_ind_houses) and domestic

hot water (i.e. therm_prod_ind_houses). Figure 1

depicts the completeness along the four periods. It can

be observed as the two initial periods cover 100% of

Figure 1: Analysis of data completeness during the 4 reports

for the intervention of hybrid solar power system in Nantes.

Analysis of Data Quality in Digital Smart Cities: The Cases of Nantes, Hamburg and Helsinki

355

samples, while third period reduces the electricity

production to 83% and fourth period is empty. This

picture highlights the requirements in terms of

surveillance systems to generate alarms for avoiding

data losses as it is the case.

In terms of correctness, energy monitoring

production relies on energy meters, which, as

explained before, are cumulative and it should be only

checked that they are not negative. In this perspective,

Figure 2 represents the maximum values for each of the

periods (except the fourth one that does not contain any

data sample). In dashed black, the maximum expected

production for electricity and, in dashed pink, the

maximum expected thermal production. It can be

extracted that electricity production is less than the

expected maximum value; therefore, working as

expected. Nevertheless, thermal production exceeds

the maximum expected value in report #3.

Figure 2: Data range of the hybrid power system variables

in Nantes.

Another example is related to PV (Photovoltaics)

plants and displayed in Figure 3, which is very

common in smart cities data collection. The first

report is empty due to the delays in the intervention,

but progressively, data completeness increased,

reaching 100% in the fourth report. This is the typical

schema, where first years entail the commissioning of

the monitoring systems (SmartCities Marketplace,

2018).

Figure 3: Data completeness analysis during the 4 reports

for the intervention of PV plant in Nantes.

4.2 Hamburg

Starting with the intervention related to the hydrogen-

based district heating, it can be observed that the

behaviour is similar to the PV plant in Nantes. As

already explained, according to the Smart Cities

Marketplace (SmartCities Marketplace, 2018), first

years are focused on the calibration of the sensors and

systems, which is perfectly observed in Figure 4. The

first report only contains 26% of the data samples, but

second and third reports increase up to 96% and 99%,

respectively. The fourth report is not documented;

therefore, no statistical analysis is available.

Figure 4: Data completeness analysis of the data-set for the

hydrogen-based district heating in Hamburg.

Due to the nature of the meter for the hydrogen-

based district heating, i.e. cumulative values,

extracting the maximum values range is not valuable.

That is to say, cumulative meters provide appended

values, i.e. summing the new value over the previous

one and instantaneous energy values can be only

obtained by subtracting consecutive samples. Then,

these measurements can reach infinite values and that

is the reason why the analysis of correctness,

complementary to the completeness, of this specific

action is not included here. In this line, an interesting

case is the one of the PV on roofs. This intervention

comes from previous years and just integrated data

into the digital urban platform. That is the reason why

100% completeness is achieved along the four

reports. However, in terms of maximum values, as

depicted in Figure 5, all the reports exceed the

maximum value for the generated electrical energy of

the PV. The reason is a slight change in the

configuration parameters than those from the original

design, allowing the detection of this misalignments

or mismatching. In other words, without the

application of this quality assessment procedure, this

error probably would have never been detected.

DATA 2022 - 11th International Conference on Data Science, Technology and Applications

356

Figure 5: Data range of the PV on roofs electricity

generation in Hamburg.

4.3 Helsinki

The first selected action to illustrate the data quality

in Helsinki is the Viikki environmental house. It

consists of five variables, as observed in Figure 6. It

is worth to mention this Viikki house stated its

monitoring prior to the project, while new energy

management strategies and geothermal pumps are

part of the mySMARTLife context. Moreover, this

building is considered as one of the most energy

efficient office buildings in Helsinki. These are the

reasons why there are very high values since the

beginning of the data collection for the total

electricity and thermal consumption, as well as PV

production. However, it is a good example to

demonstrate that there are no error-free data

collection approaches. The new variables introduced

within the project refer to the inlet and outlet

temperatures of the geothermal system. As expected,

similar to previous examples, data completeness is

higher each report, but the last report reduces the

percentage of completeness due to communication

errors in data transmission.

Figure 6: Data completeness analysis of the data-set for the

Viiki environmental house in Helsinki.

Figure 7: Data range of the Kalasatama high-performance

district in Helsinki.

In terms of correctness, the Kalasatama high-

performance district intervention, illustrated in Figure

7, is selected and shows the case of out of range

values, but these cannot be considered low quality. As

it has been explained in the methodology, maximum

and minimum values are expected according to the

experience. In this specific case, the building demand

is known, but the effects of COVID-19 are

highlighted. Figure 7 depicts the exceed of the

building heating energy consumption during 3

and

periods, when the Nordic COVID-19 strategy

encouraged working at home when possible. This

implies the increase of energy use to achieve comfort

along the entire day, incrementing the required

thermal energy. It is then clear that these values

cannot be classified as low-quality values, but

abnormal, which is the main objective of the data

quality approach. That is to say, not only discarding

data-set, but finding evidences of non-expected

behaviours as the case of Kalasatama.

5 DISCUSSION

Data quality is pivotal to make accurate decisions and

calculate KPIs when evaluating performance of city

interventions. Data quality methodologies are

developed with high interest in the research field.

However, real status of the data quality should be

investigated. This is the case of the analysis

performed in this paper in the three cities of Nantes,

Hamburg and Helsinki. In this line, the need to put

efforts in better data collection approaches should be

remarked.

According to the Smart Cities marketplace

monitoring guide (SmartCities Marketplace, 2018),

real data quality is not reached until one year and a

half have passed since the end of the interventions.

Figure 8 draws the stages that are set along the time.

Analysis of Data Quality in Digital Smart Cities: The Cases of Nantes, Hamburg and Helsinki

357

While year 0 is considered as the finalisation of the

interventions, year 1 is related to the commissioning,

when data collection is being polished and errors or

bugs are being corrected. From the beginning of the

year 2 to mid-year, the optimisation of the data

gathering process is conducted and, since second year

and a half, the optimal operation is expected.

Figure 8: Smart Cities marketplace monitoring guide.

This is exactly the case of mySMARTLife project.

As illustrated in Figure 9, Figure 10 and Figure 11,

the increase of quality from first year to second year

is notable. In Nantes, it can be observed as the second

year (dark blue) for the interventions contributes

more than the 50% of the quality, while first year

quality is very limited. Hamburg shows something

similar, although, in this case, the increment has been

lower (i.e. better performance during first year). In

contrast, Helsinki offers similar numbers during both

years analised, mainly due to the reason that many

actions were monitored before mySMARTLife.

Figure 9: Comparison of the data completeness during years

1 and 2 for the interventions in the city of Nantes.

Figure 10: Comparison of the data completeness during

years 1 and 2 for the interventions in the city of Hamburg.

Figure 11: Comparison of the data completeness during

years 1 and 2 for the interventions in the city of Helsinki.

Finally, as introduced in the methodology, a three-

range assessment is made for the data completeness

of the three cities. Table 2 collects the results for

Nantes actions (numbered) during the first year

(reports R1 and R2) and a half (third report R3),

which evidences the previous sentence about the

increase of data quality along first year. Not all the

actions are included, discarding those with delays

and; therefore, not reported.

Table 2: Analysis of the completeness according to the

established ranges in the methodology for the Nantes

actions during the three first reports (R1, R2 and R3).

Action Data sample R1 R2 R3

A1 Ener_dem_DH 100 100 100

A12

Elect_prod_ind_houses

100 100 83

Therm

rod

ind

houses 100 100 100

A14 &

A27

Elect_cons_charg_stat

0 50 100

A21a Elec_pro

0 33 100

A21b

Elect

ection 100 100 100

Elect

cons

buil

0 33 100

Elect_injection 0 33 100

Elec_pro

0 33 100

A22 &

A27

Elect_bat

0 16 100

Elect

stored

batt 0 16 100

A23a

Dist

ebus 100 100 100

Ener

cons

ebus 100 100 100

Nb_pass_ebus 50 33 100

Nb_trips_ebus 100 100 100

A24

Charg_stat_uptime_ebus

0 83 100

Ener

deliv

ebus 40 42 90

char

ebus 0 83 100

DHW

cons 100 100 100

qDHW 100 100 100

Elect_cons_build_PL 0 0 0

In the case of Hamburg (

Table 3

), data

completeness observed demonstrates an increment in

specific actions. There are cases with 100% of

DATA 2022 - 11th International Conference on Data Science, Technology and Applications

358

completeness already due to monitoring starting

before the reporting. Anyway, many of the actions are

green or yellow lights during the third report (R3)

with some minor exceptions for data samples, which

also demostrates the incremental data quality in smart

cities.

Table 3: Analysis of the completeness according to the

established ranges in the methodology for the Hamburg

actions during the three first reports (R1, R2 and R3).

Action Data sample R1 R2 R3

A15 Elect_cons_pub_light 100 100 100

A16

Elect

cons

ht 100 100 100

Elect

cons

ht 100 100 100

Elect

cons

ht 100 100 100

A25

Ener_deliv_fast

100 100 100

Nb_charg_op_fast 100 100 100

Nb_diff_users 100 100 100

A28

Ener_deliv_charging

oint

90 86 86.7

ev-status 64 55 88.8

Elec_co

0 0 0

Ener

heat

use 0 0 0

A30a Ener

deliv

fleet 100 100 100

A19a-b

1.OG - R.129 -

Zulüfter - Leistun

0 0 0

PV Prod.

(

kWh

)

33 100 100

Stromzähler -

Propangas

Kältemaschine (2Q1)

- Leistung

0 0 0

A17 &

A20

Elec

rod

WT1 100 100 100

Elec

rod

WT2 100 100 100

Elec

rod

WT3 100 100 100

Elec_prod_WT4 100 100 100

Elec_prod_WT5 100 100 100

Elec

100 100 100

Elec

50 100 100

Elec

0 0 0

Elec_pro

0 0 0

Elec_pro

0 0 33

A13 &

A18

electrical_BHKW1

26 96 99.4

electrical

BHKW2 97 96 99.4

BHKW1 26 96 99.4

BHKW2 26 96 99.4

Finally, Helsinki, which already was highly

digitalised, demonstrates that more mature cities in

monitoring strategies can reach very valuable values

in terms of data quality, hence, better-informed

decisions. Table 4 is almost green with the exception

for the new data samples introduced in the project,

which follow the same trend as explained before.

Table 4: Analysis of the completeness according to the

established ranges in the methodology for the Helsinki

actions during the three first reports (R1, R2 and R3).

Action Data sample R1 R2 R3

Elec_cons

100 100 100

Ener

heat 100 100 100

Elec

cons 100 100 100

Ener

heat 100 100 100

Elec_cons

100 100 100

Ener_heat 100 100 100

Inlet_T_cool 67 87.2 96.8

Outlet

cool 67 87.2 96.8

roduction 97 99.98 99.99

A14 &

A16

DC_consumption 100 100 100

DC_pro

100 100 100

DH_consumption 100 100 100

rod 100 100 100

rod

HP 100 100 100

A16 DH

rod

HP 100 100 100

A17 &

A18

PV_production1

100 100 100

PV_production2 100 100 100

PV_production3 100 100 100

To summarize, after having received four raw

quality reports, 39% of the actions reach at this early

stage more than 12 months of high-quality data and

47% of actions report lower values for completeness

and correctness. The remaining 14% of the actions

refer to those interventions with deviations and later

starting monitoring date. Therefore, it was not

possible to report them yet during the periods of this

preliminary analysis of data quality.

6 CONCLUSIONS

Smart Cities are not only the future but the present.

Therefore, transformation plans for more liveable

spaces and more efficient cities are required.

Decisions should be made on the basis of real and

reliable data. Nevertheless, data, when available,

usually lacks of enough quality to make rationale

decisions.

On the other hand, digitalisation of cities is slowly

progressing and quality checks are not periodically

carried out. This paper aimed at assessing the real

data quality in cities, focused on the three cities of

Nantes, Hamburg and Helsinki. Methodologies are

wide and diverse, but these are not being

implemented properly. In this sense, the major lesson

learnt is the necessity of establishing the grounds

since design. The mySMARTLife project already

considered data quality when defining the open

Analysis of Data Quality in Digital Smart Cities: The Cases of Nantes, Hamburg and Helsinki

359

specifications framework through the interoperability

mechanisms and surveillance modules.

Even though efforts have been made in the

project, this study shows that data quality procedures

should not simply be implemented, but follow-up

processes are required. Having this in mind,

mySMARTLife established 6-month periodic

analysis of data, extracting qualitative values of data

quality for two main indicators: correctness and

completeness. In terms of correctness, out of range

values allow identifying abnormal situations in the

performance of the energy systems, mobility facilities

or city infrastructures. Moreover, completeness

indicates the data gaps to provide credible and

reliable results.

The three cities demonstrate that maturity levels

in the digitalisation processes are critical. Helsinki,

more advanced in digitalisation, already reports very

high data quality indicators. Nantes and Hamburg

provided a reduced data quality in the analysis

performed, but with good values considering that the

first year of data collection usually requires

corrections and commissioning activities. After the

first year, data quality increases, leveraging data

platforms to gather raw data, obtaining information

and, thus, extracting knowledge. mySMARTLife

project is currently analysing the 4

report, although

some results have been shown along the paper.

Additionally, two additional reports are planned for

the next stages of the project. That is to say, the future

plan is to continue analysing data quality to extract

best practices in the assessment methods.

ACKNOWLEDGEMENTS

Authors would like to thank the mySMARTLife

consortium and rest of partners involved in the project

for the support. Also, the authors would like to thank

the European Commission for funding the project

under GA #731297 of the H2020 programme.

REFERENCES

EUROSTAT-a. Statistics on European cities. [online]

https://ec.europa.eu/eurostat/statistics-

explained/index.php?title=Statistics_on_European_citi

es. Accessed on 21st February 2022.

EU Research and Innovation. (2018). Final Report of the

High-Level Panel of the European Decarbonisation

Pathways Initiative; European Commission.

2030 Climate & Energy Framework. Available online:

https://ec.europa.eu/clima/policies/strategies/2030_en.

Accessed on 22nd February 2022.

Batty, M., Axhausen, K.W., Giannotti, F. et al. (2012).

Smart cities of the future. Eur. Phys. J. Spec. Top. 214,

481–518.

Hernández, J.L.; García, R.; Schonowski, J.; Atlan, D.;

Chanson, G.; Ruohomäki, T. (2020). Interoperable

Open Specifications Framework for the

Implementation of Standardized Urban Platforms.

Sensors, 20, 2402. https://doi.org/10.3390/s20082402.

mySMARTLife project. GA #731297, [online]

https://www.mysmartlife.eu/mysmartlife/, accessed on

22nd February 2022.

Quijano, A.; Hernández, J.L.; Nouaille, P.; Virtanen, M.;

Sánchez-Sarachu, B.; Pardo-Bosch, F.; Knieilng, J.

(2022). Towards Sustainable and Smart Cities:

Replicable and KPI-Driven Evaluation Framework.

Buildings, 12, 233.

Alanne, K. and Sierla, S. (2021). An overview of machine

learning applications for smart buildings, Sust. Cities

Soc., vol. 76, p. 103445.

Hasan, Z. and Roy, N. (2021). Trending machine learning

models in cyber-physical building environment: A

survey, WIREs Data Mining and Knowledge Discovery,

vol. 11, no. 5, p. e1422.

Schmidt, C.O., Struckmann, S., Enzenbach, C. et al. (2021).

Facilitating harmonized data quality assessments. A

data quality framework for observational health

research data collections with software

implementations in R. BMC Med Res Methodol 21, 63.

Sin Yong Teng, Michal Touš, Wei Dong Leong, Bing Shen

How, Hon Loong Lam, Vítězslav Máša. (2021). Recent

advances on industrial data-driven energy savings:

Digital twins and infrastructures, Renewable and

Sustainable Energy Reviews, Vol. 135, 110208.

Hossein Motlagh, N.; Mohammadrezaei, M.; Hunt, J.;

Zakeri, B. (2020). Internet of Things (IoT) and the

Energy Sector. Energies, 13, 494.

L. Ender, (2021). Data Governance in Digital Platforms: A

case analysis in the building sector, Dissertation.

Abraham, R., Schneider, J., Vom Brocke, J. (2019). Data

governance: A conceptual framework, structured

review, and research agenda. International Journal of

Information Management, 49, pp. 424-438.

Open Geospatial Consortium. (2019). OGC SensorThings

API Part 1 Sensing Version 1.1.

Araújo, T. B., Cappiello, C., Kozievitch, N. P., Mestre, D.

G., Pires, C. E. S., & Vitali, M. (2017). Towards reliable

data analyses for smart cities. In Proceedings of the 21st

International Database Engineering & Applications

Symposium (pp. 304-308).

SmartCities Marketplace. (2018). Technical Monitoring

Guide. Technical report. European Commission.

DATA 2022 - 11th International Conference on Data Science, Technology and Applications

360