SlideShare a Scribd company logo
1 of 26
Download to read offline
Smart buildings
Cameron Roach
9 November 2017
Overview
Smart buildings
Analysing the data
Building an end product
·
·
·
2/26
Smart buildings
Trying to be smart with commercial building data
How do commercial buildings work?
Facility managers (FMs) oversee the day to day operations of a commercial
building.
Used to be a whole team!
Shrinking maintenance budgets and increasing complexity makes this a
challenging problem.
Building operations are automated by Building Management Systems (BMSs).
·
·
·
·
4/26
What is a BMS?
Commercial buildings contain Building Management Systems (BMSs) to
improve indoor environment quality and reduce energy consumption.
A BMS will control heating, cooling, ventilation and lighting systems.
Contain thousands of points for sensors (temperature, humidity), actuators
(fans, motors, dampers) and software (schedule, trend logs, calculations).
A BMS will monitor sensors and adjust actuators based on their readings.
For example, if high temperatures are recorded in a room, dampers will open
and air handlers will modulate to provide cooler air.
·
·
·
·
·
5/26
6/26
How does this work in practice?
Vendor sets up a BMS. The BMS will behave in a certain way based predefined
rules.
BMS systems are costly to implement and to modify. Can require a lot of
coding to change the BMS's behaviour.
The bigger the BMS is the harder it is to find what matters. Locating problems
is difficult and time-consuming.
For example, a heating valve might be locked open. If this isn't detected the
BMS will cool the room to reach the required temperature.
·
·
·
·
7/26
So what can we do?
Help facility managers identify if a BMS is operating optimally.
Buildings Alive's goal is to
·
Fault detection
Diagnostics
-
-
·
Collect BMS data using our E2 device
Analyse and transform data into useful information.
Help guide FM's to find out what's wrong.
Provide timely and actionable information.
-
-
-
-
8/26
Analysing the data
Feature generation, dimensionality reduction, clustering
10/26
11/26
Feature generation
Dealing with thousands of unevenly spaced time-series.
Uneven spacing in time-series presents difficulties.
Rather than rounding or imputing data we can generate features and work
with them instead.
·
·
·
12/26
What features might be useful?
Feature generation for time-series clustering is discussed in Wang, Smith, and Hyndman (2006). Some
useful features for our case might be
Normalise these features using their median, , and interquartile range, ,
Mean
Standard deviation
Kurtosis
Skewness
Biggest change ( )
Smallest change ( )
Number of "mean crossings" per day
·
·
·
·
· { − }maxi
∣
∣yti
yti−1
∣
∣
· { − }mini
∣
∣yti
yti−1
∣
∣
·
M IQR
= .y
∗
y − M
IQR
13/26
14/26
15/26
Dimension reduction and clustering
Too many sensors to visualise easily.
Use dimensionality reduction.
Identify clusters and singletons.
·
·
·
16/26
Which clustering algorithm?
Method Advantages Disadvantages
K-means Easy to learn. Outperformed by other algorithms.
Hierarchical clustering
Informative - produces a
dendrogram.
Not suitable for large data sets -
time complexity.
Affinity propagation
Automatically determines number of
clusters.
Not suitable for large data sets - time
complexity.
Spectral clustering Good performance.
See Nadler and Galun (2007). Time complexity
of .
( log(n))n
2
( t)n
2
( )n
3
17/26
Image: (“Comparing Different Clustering Algorithms on Toy Datasets” 2017)
18/26
Obligatory mathematics slide
Spectral clustering
We are given points and a similarity matrix . Define the weight matrix, degree matrix and
graph Laplacian as
where,
Once is determined find the eigenvectors corresponding to the smallest eigenvalues of .
Finally, cluster the rows of using K-means.
n ∈xi ℝ
p
S
W
D
L
= ( ) ∈wij ℝ
n×n
= diag ( )di
= D − W,
is the weight between nodes and based on , and,
is the weighted degree of node .
· wij i j S
· =di ∑
n
j=1
wij i
L m Zn×m m L
Zn×m
19/26
Building a prototype
Prototyping with Dash
Dash
Recently released by Plotly.
Easily build web applications for
data analytics.
Open sourced under the MIT
license.
Works nicely with the existing Plotly
graphing libraries.
·
·
·
·
Python equivalent of R's Shiny.·
21/26
22/26
Simple example
import dash
from dash.dependencies import Input, Output
import dash_core_components as dcc
import dash_html_components as html
import pandas as pd
app = dash.Dash()
app.layout = html.Div([
dcc.Dropdown(id='my-dropdown',
options=[{'label': 'Option A', 'value': 'A'},
{'label': 'Option B', 'value': 'B'}]),
dcc.Graph(id='my-graph')
])
@app.callback(Output('my-graph', 'figure'),
[Input('my-dropdown', 'value')])
def update_graph(dd_value):
df_query = df.query("Variable == @dd_value")
return {'data': [{'x': df_query.x, 'y': df_query.y}]}
if __name__ == '__main__':
app.run_server()
23/26
Demonstration
24/26
Thank you for listening!
Any questions?
References
“Comparing Different Clustering Algorithms on Toy Datasets.” 2017. http://scikit-
learn.org/stable/auto_examples/cluster/plot_cluster_comparison.html.
Friedman, Jerome, Trevor Hastie, and Robert Tibshirani. 2001. The Elements of
Statistical Learning. Vol. 1. Springer series in statistics New York.
Murphy, Kevin P. 2012. Machine Learning: A Probabilistic Perspective. MIT Press.
Nadler, Boaz, and Meirav Galun. 2007. “Fundamental Limitations of Spectral
Clustering.” In Advances in Neural Information Processing Systems 19, edited by P B
Schölkopf, J C Platt, and T Hoffman, 1017–24. MIT Press.
Von Luxburg, Ulrike. 2007. “A Tutorial on Spectral Clustering.” Statistics and
Computing.
Wang, Xiaozhe, Kate Smith, and Rob Hyndman. 2006. “Characteristic-Based
Clustering for Time Series Data.” Data Mining and Knowledge Discovery 13 (3): 335–
64.
26/26

More Related Content

Similar to Smart buildings

An Introduction to Cloud Computing by Robert Grossman 08-06-09 (v19)
An Introduction to Cloud Computing by Robert Grossman 08-06-09 (v19)An Introduction to Cloud Computing by Robert Grossman 08-06-09 (v19)
An Introduction to Cloud Computing by Robert Grossman 08-06-09 (v19)Robert Grossman
 
The Impact of Cloud Computing on Predictive Analytics 7-29-09 v5
The Impact of Cloud Computing on Predictive Analytics 7-29-09 v5The Impact of Cloud Computing on Predictive Analytics 7-29-09 v5
The Impact of Cloud Computing on Predictive Analytics 7-29-09 v5Robert Grossman
 
Apache SystemML Optimizer and Runtime techniques by Matthias Boehm
Apache SystemML Optimizer and Runtime techniques by Matthias BoehmApache SystemML Optimizer and Runtime techniques by Matthias Boehm
Apache SystemML Optimizer and Runtime techniques by Matthias BoehmArvind Surve
 
Apache SystemML Optimizer and Runtime techniques by Matthias Boehm
Apache SystemML Optimizer and Runtime techniques by Matthias BoehmApache SystemML Optimizer and Runtime techniques by Matthias Boehm
Apache SystemML Optimizer and Runtime techniques by Matthias BoehmArvind Surve
 
Project A Data Modelling Best Practices Part II: How to Build a Data Warehouse?
Project A Data Modelling Best Practices Part II: How to Build a Data Warehouse?Project A Data Modelling Best Practices Part II: How to Build a Data Warehouse?
Project A Data Modelling Best Practices Part II: How to Build a Data Warehouse?Martin Loetzsch
 
Ijariie1117 volume 1-issue 1-page-25-27
Ijariie1117 volume 1-issue 1-page-25-27Ijariie1117 volume 1-issue 1-page-25-27
Ijariie1117 volume 1-issue 1-page-25-27IJARIIE JOURNAL
 
IRJET- Load Balancing and Crash Management in IoT Environment
IRJET-  	  Load Balancing and Crash Management in IoT EnvironmentIRJET-  	  Load Balancing and Crash Management in IoT Environment
IRJET- Load Balancing and Crash Management in IoT EnvironmentIRJET Journal
 
Dask glm-scipy2017-final
Dask glm-scipy2017-finalDask glm-scipy2017-final
Dask glm-scipy2017-finalHussain Sultan
 
Comparative study of optimization algorithms on convolutional network for aut...
Comparative study of optimization algorithms on convolutional network for aut...Comparative study of optimization algorithms on convolutional network for aut...
Comparative study of optimization algorithms on convolutional network for aut...IJECEIAES
 
CoolSim Overview
CoolSim OverviewCoolSim Overview
CoolSim Overviewpaulbemis
 
A Parallel Data Distribution Management Algorithm
A Parallel Data Distribution Management AlgorithmA Parallel Data Distribution Management Algorithm
A Parallel Data Distribution Management AlgorithmGabriele D'Angelo
 
Kaggle Winning Solution Xgboost algorithm -- Let us learn from its author
Kaggle Winning Solution Xgboost algorithm -- Let us learn from its authorKaggle Winning Solution Xgboost algorithm -- Let us learn from its author
Kaggle Winning Solution Xgboost algorithm -- Let us learn from its authorVivian S. Zhang
 
The GREDOR project. Redesigning the decision chain for managing distribution ...
The GREDOR project. Redesigning the decision chain for managing distribution ...The GREDOR project. Redesigning the decision chain for managing distribution ...
The GREDOR project. Redesigning the decision chain for managing distribution ...Université de Liège (ULg)
 

Similar to Smart buildings (20)

An Introduction to Cloud Computing by Robert Grossman 08-06-09 (v19)
An Introduction to Cloud Computing by Robert Grossman 08-06-09 (v19)An Introduction to Cloud Computing by Robert Grossman 08-06-09 (v19)
An Introduction to Cloud Computing by Robert Grossman 08-06-09 (v19)
 
The Impact of Cloud Computing on Predictive Analytics 7-29-09 v5
The Impact of Cloud Computing on Predictive Analytics 7-29-09 v5The Impact of Cloud Computing on Predictive Analytics 7-29-09 v5
The Impact of Cloud Computing on Predictive Analytics 7-29-09 v5
 
autoTVM
autoTVMautoTVM
autoTVM
 
GREDOR
GREDORGREDOR
GREDOR
 
Apache SystemML Optimizer and Runtime techniques by Matthias Boehm
Apache SystemML Optimizer and Runtime techniques by Matthias BoehmApache SystemML Optimizer and Runtime techniques by Matthias Boehm
Apache SystemML Optimizer and Runtime techniques by Matthias Boehm
 
Apache SystemML Optimizer and Runtime techniques by Matthias Boehm
Apache SystemML Optimizer and Runtime techniques by Matthias BoehmApache SystemML Optimizer and Runtime techniques by Matthias Boehm
Apache SystemML Optimizer and Runtime techniques by Matthias Boehm
 
Project A Data Modelling Best Practices Part II: How to Build a Data Warehouse?
Project A Data Modelling Best Practices Part II: How to Build a Data Warehouse?Project A Data Modelling Best Practices Part II: How to Build a Data Warehouse?
Project A Data Modelling Best Practices Part II: How to Build a Data Warehouse?
 
A detailed analysis of the supervised machine Learning Algorithms
A detailed analysis of the supervised machine Learning AlgorithmsA detailed analysis of the supervised machine Learning Algorithms
A detailed analysis of the supervised machine Learning Algorithms
 
Ijariie1117 volume 1-issue 1-page-25-27
Ijariie1117 volume 1-issue 1-page-25-27Ijariie1117 volume 1-issue 1-page-25-27
Ijariie1117 volume 1-issue 1-page-25-27
 
IRJET- Load Balancing and Crash Management in IoT Environment
IRJET-  	  Load Balancing and Crash Management in IoT EnvironmentIRJET-  	  Load Balancing and Crash Management in IoT Environment
IRJET- Load Balancing and Crash Management in IoT Environment
 
Dask glm-scipy2017-final
Dask glm-scipy2017-finalDask glm-scipy2017-final
Dask glm-scipy2017-final
 
Comparative study of optimization algorithms on convolutional network for aut...
Comparative study of optimization algorithms on convolutional network for aut...Comparative study of optimization algorithms on convolutional network for aut...
Comparative study of optimization algorithms on convolutional network for aut...
 
CoolSim Overview
CoolSim OverviewCoolSim Overview
CoolSim Overview
 
Lec1
Lec1Lec1
Lec1
 
A Parallel Data Distribution Management Algorithm
A Parallel Data Distribution Management AlgorithmA Parallel Data Distribution Management Algorithm
A Parallel Data Distribution Management Algorithm
 
Kaggle Winning Solution Xgboost algorithm -- Let us learn from its author
Kaggle Winning Solution Xgboost algorithm -- Let us learn from its authorKaggle Winning Solution Xgboost algorithm -- Let us learn from its author
Kaggle Winning Solution Xgboost algorithm -- Let us learn from its author
 
The GREDOR project. Redesigning the decision chain for managing distribution ...
The GREDOR project. Redesigning the decision chain for managing distribution ...The GREDOR project. Redesigning the decision chain for managing distribution ...
The GREDOR project. Redesigning the decision chain for managing distribution ...
 
A04230105
A04230105A04230105
A04230105
 
4.2_Microgrid Design Toolkit_Eddy_EPRI/SNL Microgrid
4.2_Microgrid Design Toolkit_Eddy_EPRI/SNL Microgrid4.2_Microgrid Design Toolkit_Eddy_EPRI/SNL Microgrid
4.2_Microgrid Design Toolkit_Eddy_EPRI/SNL Microgrid
 
Cloud Computing
Cloud ComputingCloud Computing
Cloud Computing
 

Recently uploaded

Hubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroidsHubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroidsSérgio Sacani
 
Pests of cotton_Sucking_Pests_Dr.UPR.pdf
Pests of cotton_Sucking_Pests_Dr.UPR.pdfPests of cotton_Sucking_Pests_Dr.UPR.pdf
Pests of cotton_Sucking_Pests_Dr.UPR.pdfPirithiRaju
 
Botany krishna series 2nd semester Only Mcq type questions
Botany krishna series 2nd semester Only Mcq type questionsBotany krishna series 2nd semester Only Mcq type questions
Botany krishna series 2nd semester Only Mcq type questionsSumit Kumar yadav
 
Recombination DNA Technology (Nucleic Acid Hybridization )
Recombination DNA Technology (Nucleic Acid Hybridization )Recombination DNA Technology (Nucleic Acid Hybridization )
Recombination DNA Technology (Nucleic Acid Hybridization )aarthirajkumar25
 
Biological Classification BioHack (3).pdf
Biological Classification BioHack (3).pdfBiological Classification BioHack (3).pdf
Biological Classification BioHack (3).pdfmuntazimhurra
 
Nightside clouds and disequilibrium chemistry on the hot Jupiter WASP-43b
Nightside clouds and disequilibrium chemistry on the hot Jupiter WASP-43bNightside clouds and disequilibrium chemistry on the hot Jupiter WASP-43b
Nightside clouds and disequilibrium chemistry on the hot Jupiter WASP-43bSérgio Sacani
 
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptxSOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptxkessiyaTpeter
 
Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?Patrick Diehl
 
Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |
Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |
Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |aasikanpl
 
Broad bean, Lima Bean, Jack bean, Ullucus.pptx
Broad bean, Lima Bean, Jack bean, Ullucus.pptxBroad bean, Lima Bean, Jack bean, Ullucus.pptx
Broad bean, Lima Bean, Jack bean, Ullucus.pptxjana861314
 
Spermiogenesis or Spermateleosis or metamorphosis of spermatid
Spermiogenesis or Spermateleosis or metamorphosis of spermatidSpermiogenesis or Spermateleosis or metamorphosis of spermatid
Spermiogenesis or Spermateleosis or metamorphosis of spermatidSarthak Sekhar Mondal
 
Animal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptxAnimal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptxUmerFayaz5
 
Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...
Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...
Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...jana861314
 
Formation of low mass protostars and their circumstellar disks
Formation of low mass protostars and their circumstellar disksFormation of low mass protostars and their circumstellar disks
Formation of low mass protostars and their circumstellar disksSérgio Sacani
 
Natural Polymer Based Nanomaterials
Natural Polymer Based NanomaterialsNatural Polymer Based Nanomaterials
Natural Polymer Based NanomaterialsAArockiyaNisha
 
Lucknow 💋 Russian Call Girls Lucknow Finest Escorts Service 8923113531 Availa...
Lucknow 💋 Russian Call Girls Lucknow Finest Escorts Service 8923113531 Availa...Lucknow 💋 Russian Call Girls Lucknow Finest Escorts Service 8923113531 Availa...
Lucknow 💋 Russian Call Girls Lucknow Finest Escorts Service 8923113531 Availa...anilsa9823
 
Recombinant DNA technology (Immunological screening)
Recombinant DNA technology (Immunological screening)Recombinant DNA technology (Immunological screening)
Recombinant DNA technology (Immunological screening)PraveenaKalaiselvan1
 
Presentation Vikram Lander by Vedansh Gupta.pptx
Presentation Vikram Lander by Vedansh Gupta.pptxPresentation Vikram Lander by Vedansh Gupta.pptx
Presentation Vikram Lander by Vedansh Gupta.pptxgindu3009
 
A relative description on Sonoporation.pdf
A relative description on Sonoporation.pdfA relative description on Sonoporation.pdf
A relative description on Sonoporation.pdfnehabiju2046
 
Nanoparticles synthesis and characterization​ ​
Nanoparticles synthesis and characterization​  ​Nanoparticles synthesis and characterization​  ​
Nanoparticles synthesis and characterization​ ​kaibalyasahoo82800
 

Recently uploaded (20)

Hubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroidsHubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroids
 
Pests of cotton_Sucking_Pests_Dr.UPR.pdf
Pests of cotton_Sucking_Pests_Dr.UPR.pdfPests of cotton_Sucking_Pests_Dr.UPR.pdf
Pests of cotton_Sucking_Pests_Dr.UPR.pdf
 
Botany krishna series 2nd semester Only Mcq type questions
Botany krishna series 2nd semester Only Mcq type questionsBotany krishna series 2nd semester Only Mcq type questions
Botany krishna series 2nd semester Only Mcq type questions
 
Recombination DNA Technology (Nucleic Acid Hybridization )
Recombination DNA Technology (Nucleic Acid Hybridization )Recombination DNA Technology (Nucleic Acid Hybridization )
Recombination DNA Technology (Nucleic Acid Hybridization )
 
Biological Classification BioHack (3).pdf
Biological Classification BioHack (3).pdfBiological Classification BioHack (3).pdf
Biological Classification BioHack (3).pdf
 
Nightside clouds and disequilibrium chemistry on the hot Jupiter WASP-43b
Nightside clouds and disequilibrium chemistry on the hot Jupiter WASP-43bNightside clouds and disequilibrium chemistry on the hot Jupiter WASP-43b
Nightside clouds and disequilibrium chemistry on the hot Jupiter WASP-43b
 
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptxSOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
 
Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?
 
Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |
Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |
Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |
 
Broad bean, Lima Bean, Jack bean, Ullucus.pptx
Broad bean, Lima Bean, Jack bean, Ullucus.pptxBroad bean, Lima Bean, Jack bean, Ullucus.pptx
Broad bean, Lima Bean, Jack bean, Ullucus.pptx
 
Spermiogenesis or Spermateleosis or metamorphosis of spermatid
Spermiogenesis or Spermateleosis or metamorphosis of spermatidSpermiogenesis or Spermateleosis or metamorphosis of spermatid
Spermiogenesis or Spermateleosis or metamorphosis of spermatid
 
Animal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptxAnimal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptx
 
Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...
Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...
Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...
 
Formation of low mass protostars and their circumstellar disks
Formation of low mass protostars and their circumstellar disksFormation of low mass protostars and their circumstellar disks
Formation of low mass protostars and their circumstellar disks
 
Natural Polymer Based Nanomaterials
Natural Polymer Based NanomaterialsNatural Polymer Based Nanomaterials
Natural Polymer Based Nanomaterials
 
Lucknow 💋 Russian Call Girls Lucknow Finest Escorts Service 8923113531 Availa...
Lucknow 💋 Russian Call Girls Lucknow Finest Escorts Service 8923113531 Availa...Lucknow 💋 Russian Call Girls Lucknow Finest Escorts Service 8923113531 Availa...
Lucknow 💋 Russian Call Girls Lucknow Finest Escorts Service 8923113531 Availa...
 
Recombinant DNA technology (Immunological screening)
Recombinant DNA technology (Immunological screening)Recombinant DNA technology (Immunological screening)
Recombinant DNA technology (Immunological screening)
 
Presentation Vikram Lander by Vedansh Gupta.pptx
Presentation Vikram Lander by Vedansh Gupta.pptxPresentation Vikram Lander by Vedansh Gupta.pptx
Presentation Vikram Lander by Vedansh Gupta.pptx
 
A relative description on Sonoporation.pdf
A relative description on Sonoporation.pdfA relative description on Sonoporation.pdf
A relative description on Sonoporation.pdf
 
Nanoparticles synthesis and characterization​ ​
Nanoparticles synthesis and characterization​  ​Nanoparticles synthesis and characterization​  ​
Nanoparticles synthesis and characterization​ ​
 

Smart buildings

  • 2. Overview Smart buildings Analysing the data Building an end product · · · 2/26
  • 3. Smart buildings Trying to be smart with commercial building data
  • 4. How do commercial buildings work? Facility managers (FMs) oversee the day to day operations of a commercial building. Used to be a whole team! Shrinking maintenance budgets and increasing complexity makes this a challenging problem. Building operations are automated by Building Management Systems (BMSs). · · · · 4/26
  • 5. What is a BMS? Commercial buildings contain Building Management Systems (BMSs) to improve indoor environment quality and reduce energy consumption. A BMS will control heating, cooling, ventilation and lighting systems. Contain thousands of points for sensors (temperature, humidity), actuators (fans, motors, dampers) and software (schedule, trend logs, calculations). A BMS will monitor sensors and adjust actuators based on their readings. For example, if high temperatures are recorded in a room, dampers will open and air handlers will modulate to provide cooler air. · · · · · 5/26
  • 7. How does this work in practice? Vendor sets up a BMS. The BMS will behave in a certain way based predefined rules. BMS systems are costly to implement and to modify. Can require a lot of coding to change the BMS's behaviour. The bigger the BMS is the harder it is to find what matters. Locating problems is difficult and time-consuming. For example, a heating valve might be locked open. If this isn't detected the BMS will cool the room to reach the required temperature. · · · · 7/26
  • 8. So what can we do? Help facility managers identify if a BMS is operating optimally. Buildings Alive's goal is to · Fault detection Diagnostics - - · Collect BMS data using our E2 device Analyse and transform data into useful information. Help guide FM's to find out what's wrong. Provide timely and actionable information. - - - - 8/26
  • 9. Analysing the data Feature generation, dimensionality reduction, clustering
  • 10. 10/26
  • 11. 11/26
  • 12. Feature generation Dealing with thousands of unevenly spaced time-series. Uneven spacing in time-series presents difficulties. Rather than rounding or imputing data we can generate features and work with them instead. · · · 12/26
  • 13. What features might be useful? Feature generation for time-series clustering is discussed in Wang, Smith, and Hyndman (2006). Some useful features for our case might be Normalise these features using their median, , and interquartile range, , Mean Standard deviation Kurtosis Skewness Biggest change ( ) Smallest change ( ) Number of "mean crossings" per day · · · · · { − }maxi ∣ ∣yti yti−1 ∣ ∣ · { − }mini ∣ ∣yti yti−1 ∣ ∣ · M IQR = .y ∗ y − M IQR 13/26
  • 14. 14/26
  • 15. 15/26
  • 16. Dimension reduction and clustering Too many sensors to visualise easily. Use dimensionality reduction. Identify clusters and singletons. · · · 16/26
  • 17. Which clustering algorithm? Method Advantages Disadvantages K-means Easy to learn. Outperformed by other algorithms. Hierarchical clustering Informative - produces a dendrogram. Not suitable for large data sets - time complexity. Affinity propagation Automatically determines number of clusters. Not suitable for large data sets - time complexity. Spectral clustering Good performance. See Nadler and Galun (2007). Time complexity of . ( log(n))n 2 ( t)n 2 ( )n 3 17/26
  • 18. Image: (“Comparing Different Clustering Algorithms on Toy Datasets” 2017) 18/26
  • 19. Obligatory mathematics slide Spectral clustering We are given points and a similarity matrix . Define the weight matrix, degree matrix and graph Laplacian as where, Once is determined find the eigenvectors corresponding to the smallest eigenvalues of . Finally, cluster the rows of using K-means. n ∈xi ℝ p S W D L = ( ) ∈wij ℝ n×n = diag ( )di = D − W, is the weight between nodes and based on , and, is the weighted degree of node . · wij i j S · =di ∑ n j=1 wij i L m Zn×m m L Zn×m 19/26
  • 21. Dash Recently released by Plotly. Easily build web applications for data analytics. Open sourced under the MIT license. Works nicely with the existing Plotly graphing libraries. · · · · Python equivalent of R's Shiny.· 21/26
  • 22. 22/26
  • 23. Simple example import dash from dash.dependencies import Input, Output import dash_core_components as dcc import dash_html_components as html import pandas as pd app = dash.Dash() app.layout = html.Div([ dcc.Dropdown(id='my-dropdown', options=[{'label': 'Option A', 'value': 'A'}, {'label': 'Option B', 'value': 'B'}]), dcc.Graph(id='my-graph') ]) @app.callback(Output('my-graph', 'figure'), [Input('my-dropdown', 'value')]) def update_graph(dd_value): df_query = df.query("Variable == @dd_value") return {'data': [{'x': df_query.x, 'y': df_query.y}]} if __name__ == '__main__': app.run_server() 23/26
  • 25. Thank you for listening! Any questions?
  • 26. References “Comparing Different Clustering Algorithms on Toy Datasets.” 2017. http://scikit- learn.org/stable/auto_examples/cluster/plot_cluster_comparison.html. Friedman, Jerome, Trevor Hastie, and Robert Tibshirani. 2001. The Elements of Statistical Learning. Vol. 1. Springer series in statistics New York. Murphy, Kevin P. 2012. Machine Learning: A Probabilistic Perspective. MIT Press. Nadler, Boaz, and Meirav Galun. 2007. “Fundamental Limitations of Spectral Clustering.” In Advances in Neural Information Processing Systems 19, edited by P B Schölkopf, J C Platt, and T Hoffman, 1017–24. MIT Press. Von Luxburg, Ulrike. 2007. “A Tutorial on Spectral Clustering.” Statistics and Computing. Wang, Xiaozhe, Kate Smith, and Rob Hyndman. 2006. “Characteristic-Based Clustering for Time Series Data.” Data Mining and Knowledge Discovery 13 (3): 335– 64. 26/26