LSTM network time series predicts high-risk network time series predicts hig¢  LSTM...

download LSTM network time series predicts high-risk network time series predicts hig¢  LSTM network time series

of 10

  • date post

    22-Sep-2020
  • Category

    Documents

  • view

    0
  • download

    0

Embed Size (px)

Transcript of LSTM network time series predicts high-risk network time series predicts hig¢  LSTM...

  • See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/326838388

    LSTM network time series predicts high-risk tenants

    Presentation · July 2018

    DOI: 10.13140/RG.2.2.19849.95849

    CITATIONS

    0 READS

    5

    6 authors, including:

    Some of the authors of this publication are also working on these related projects:

    One Small Step for Machine, One Giant Leap for the Market: Lessons Learnt from a Real World Machine Learning Project View project

    Bus Service Optimisations View project

    Wolfgang Garn

    University of Surrey

    17 PUBLICATIONS   22 CITATIONS   

    SEE PROFILE

    All content following this page was uploaded by Wolfgang Garn on 06 August 2018.

    The user has requested enhancement of the downloaded file.

    https://www.researchgate.net/publication/326838388_LSTM_network_time_series_predicts_high-risk_tenants?enrichId=rgreq-2cedab5e8c57466da5b185d963719856-XXX&enrichSource=Y292ZXJQYWdlOzMyNjgzODM4ODtBUzo2NTY2MTYwNjgyMTg4ODRAMTUzMzU2MDg3ODk2OQ%3D%3D&el=1_x_2&_esc=publicationCoverPdf https://www.researchgate.net/publication/326838388_LSTM_network_time_series_predicts_high-risk_tenants?enrichId=rgreq-2cedab5e8c57466da5b185d963719856-XXX&enrichSource=Y292ZXJQYWdlOzMyNjgzODM4ODtBUzo2NTY2MTYwNjgyMTg4ODRAMTUzMzU2MDg3ODk2OQ%3D%3D&el=1_x_3&_esc=publicationCoverPdf https://www.researchgate.net/project/One-Small-Step-for-Machine-One-Giant-Leap-for-the-Market-Lessons-Learnt-from-a-Real-World-Machine-Learning-Project?enrichId=rgreq-2cedab5e8c57466da5b185d963719856-XXX&enrichSource=Y292ZXJQYWdlOzMyNjgzODM4ODtBUzo2NTY2MTYwNjgyMTg4ODRAMTUzMzU2MDg3ODk2OQ%3D%3D&el=1_x_9&_esc=publicationCoverPdf https://www.researchgate.net/project/Bus-Service-Optimisations?enrichId=rgreq-2cedab5e8c57466da5b185d963719856-XXX&enrichSource=Y292ZXJQYWdlOzMyNjgzODM4ODtBUzo2NTY2MTYwNjgyMTg4ODRAMTUzMzU2MDg3ODk2OQ%3D%3D&el=1_x_9&_esc=publicationCoverPdf https://www.researchgate.net/?enrichId=rgreq-2cedab5e8c57466da5b185d963719856-XXX&enrichSource=Y292ZXJQYWdlOzMyNjgzODM4ODtBUzo2NTY2MTYwNjgyMTg4ODRAMTUzMzU2MDg3ODk2OQ%3D%3D&el=1_x_1&_esc=publicationCoverPdf https://www.researchgate.net/profile/Wolfgang_Garn?enrichId=rgreq-2cedab5e8c57466da5b185d963719856-XXX&enrichSource=Y292ZXJQYWdlOzMyNjgzODM4ODtBUzo2NTY2MTYwNjgyMTg4ODRAMTUzMzU2MDg3ODk2OQ%3D%3D&el=1_x_4&_esc=publicationCoverPdf https://www.researchgate.net/profile/Wolfgang_Garn?enrichId=rgreq-2cedab5e8c57466da5b185d963719856-XXX&enrichSource=Y292ZXJQYWdlOzMyNjgzODM4ODtBUzo2NTY2MTYwNjgyMTg4ODRAMTUzMzU2MDg3ODk2OQ%3D%3D&el=1_x_5&_esc=publicationCoverPdf https://www.researchgate.net/institution/University_of_Surrey?enrichId=rgreq-2cedab5e8c57466da5b185d963719856-XXX&enrichSource=Y292ZXJQYWdlOzMyNjgzODM4ODtBUzo2NTY2MTYwNjgyMTg4ODRAMTUzMzU2MDg3ODk2OQ%3D%3D&el=1_x_6&_esc=publicationCoverPdf https://www.researchgate.net/profile/Wolfgang_Garn?enrichId=rgreq-2cedab5e8c57466da5b185d963719856-XXX&enrichSource=Y292ZXJQYWdlOzMyNjgzODM4ODtBUzo2NTY2MTYwNjgyMTg4ODRAMTUzMzU2MDg3ODk2OQ%3D%3D&el=1_x_7&_esc=publicationCoverPdf https://www.researchgate.net/profile/Wolfgang_Garn?enrichId=rgreq-2cedab5e8c57466da5b185d963719856-XXX&enrichSource=Y292ZXJQYWdlOzMyNjgzODM4ODtBUzo2NTY2MTYwNjgyMTg4ODRAMTUzMzU2MDg3ODk2OQ%3D%3D&el=1_x_10&_esc=publicationCoverPdf

  • 10/07/2018

    1

    LSTM network time series

    predicts high-risk tenants

    WOLFGANG GARN, Y IN HU, PAUL NICHOLSON, BEVAN JONES,

    HONGYING TANG, WILLIAM WRIGHT

    LSTM network time series predicts high-risk tenants Wolfgang Garn, Yin Hu, Paul Nicholson, Bevan Jones, Hongying Tang

    Overview Benefits to the real estate sector  decreasing lost revenue,  increasing efficiency  providing more help to tenants before their debt becomes

    unmanageable

    How?  Measure arrears risk for each individual tenant  differentiate between short-term and long-term arrears risk  predict trajectory of arrears  Identify factors that can be operationally used to assist tenants

    arrears management

    Structure  Background  Data - arrears Risk & Factors  Method - LSTM – network  Results & Insights

    Abstract In the United Kingdom, local councils and housing associations provide social housing at secure, low-rent housing options to those most in need. Occasionally many tenants have difficulties in paying their rent on time and fall into arrears. The lost revenue can cause substantial financial burden for these agents, while falling into arrears can cause stress to tenants. An efficient arrears management scheme is to target those who are more at risk of falling into long-term arrears so that interventions can be used for persistent loss of revenue. In our research, a Long Short-Term Memory Network (LSTM) based time series prediction model is implemented to differentiate the high-risk tenants from relatively temporary ones. This model measures the arrears risk to differentiate between short-term and long-term arrears risk and predicts the trajectory of arrears for each individual tenant. Moreover, further arrears analysis is conducted to investigate which factors could provide more assistance for tenants before their debt becomes unmanageable. A five-year rent arrears dataset is used to train and evaluate the proposed model. The root mean squared error (RMSE) is used to punish large errors by measuring of the differences between the arrears actually observed and arrears predicted by a model. Hence, this model brings benefits to the real estate sector by decreasing lost revenue, increasing efficiency and providing more help to tenants before their debt becomes unmanageable.

    LSTM network time series predicts high-risk tenants Wolfgang Garn, Yin Hu, Paul Nicholson, Bevan Jones, Hongying Tang

    Background Hastoe Housing Association  Creating opportunities to innovate

    Sustainable Homes Ltd  Influencing policy and practice

    University of Surrey - Computing and Business School

    The associate

    Innovate UK funding for 3 years

    Knowledge Transfer Partnership

  • 10/07/2018

    2

    LSTM network time series predicts high-risk tenants Wolfgang Garn, Yin Hu, Paul Nicholson, Bevan Jones, Hongying Tang

    Content

    Background Data – rent arrears & factors Method - LSTM – network Results & Insights

    LSTM network time series predicts high-risk tenants Wolfgang Garn, Yin Hu, Paul Nicholson, Bevan Jones, Hongying Tang

    Data sources Heterogeneous sets  Excel maps, databases, …

    Mosaic UK  Geodemographic classification of households

    Geographic's of vacant homes  Deprived areas more likely to be in rent arrears

     Finances  The rich, the poor and the rest

    Rent arrears (2555 tenants)

    Data: five years John Naisbitt wrote “We are drowning in information but starved for knowledge.”

    LSTM network time series predicts high-risk tenants Wolfgang Garn, Yin Hu, Paul Nicholson, Bevan Jones, Hongying Tang

    Scatter plots & Correlations

    Correlation analysis of housing beneficiaries does not support associations between factors

    Factors ◦ Energy Efficiency (ree)

    ◦ Housing Benefits (rhb)

    ◦ Rent in arrears (rma)

    ree -0.014 -0.004

    -0.014 rhb -0.122

    -0.004 -0.122 rma

    r values

    10/07/2018 6

    S A

    P r

    a ti n g

    H o u s in

    g b

    e n e fi t

    M o n th

    i n a

    rr e a rs

    SAP rating Housing benefit Month in arrears

    0.00

    0.02

    0.04

    0.06

    Corr:

    -0.0139

    Corr:

    -0.00416

    0

    50

    100

    150

    200

    Corr:

    -0.122

    -10

    0

    10

    40 60 80 0 50 100 150 200 -10 0 10

  • 10/07/2018

    3

    LSTM network time series predicts high-risk tenants Wolfgang Garn, Yin Hu, Paul Nicholson, Bevan Jones, Hongying Tang

    Indirect factors Benefits >> Energy efficiency >> rent arrears

    LSTM network time series predicts high-risk tenants Wolfgang Garn, Yin Hu, Paul Nicholson, Bevan Jones, Hongying Tang

    Rent arrears & Energy efficiency A detour … the whole housing market

    SAP = Standard Assessment Procedure for energy efficiency

    Surprisingly much rent arrears

    Lowest SAP rating –> highest rent arrears

    Highest SAP rating -> lowest rent arrears

    Other SAP ratings debatable

    F E D C B A

    LSTM network time series predicts high-risk tenants Wolfgang Garn, Yin Hu, Paul Nicholson, Bevan Jones, Hongying Tang

    Chase for features Rent arrears and average SAP rating

    Average at a certain level >>

    potential for systematic feature engineering?

    From a subject matter expert’s view: Rent arrears drops as the SAP rating increases

    s

  • 10/07/2018

    4

    LSTM network time series predicts high-risk tenants Wolfgang Garn, Yin Hu, Paul Nicholson, Bevan Jones, Hongying Tang

    Averages are better!? Correlations become visible

    More importantly patterns become visible Feature engineering “can start”

    LSTM network time series predicts high-risk tenants Wolfgang Garn, Yin Hu, Paul Nicholson, Bevan Jones, Hongying Tang

    Rent arrears & SAP “again” The better the property the less rent arrears

    LSTM network time series predicts high-risk tenants Wolfgang Garn, Yin Hu, Paul Nic