PRESENTATION: Introduction to data mining and analysis

  • Published on
    31-Jan-2016

  • View
    7

  • Download
    0

Embed Size (px)

DESCRIPTION

Presented by Gilbert PINOT of Universite Haute-Alsace last 27-31 October 2015 in Bogor, Indonesia at the Climate Change: Observation, Analysis and Health Conference

Transcript

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 1

    Introduction to

    data mining and analysis

    Gilbert PINOT

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 2

    Big Data & Data Mining

    2003 : big bang of digital stored informations

    10 21

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 3

    Big Data & Data Mining huge amount of data created daily

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 4

    Big Data & Data Mining data natures

    textetexte texte texte textetexte texte texte textetexte textetexte te

    xte

    im

    ag

    e

    au

    dio

    vid

    eo

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 5

    Big Data & Data Mining data sources

    .

    .

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 6

    usual data based systems

    high pertinence data

    DATA

    DATA

    DATA data data

    data data data

    data

    data data

    data data

    data data data

    data data

    data data

    data

    data

    data

    data data

    low pertinence data

    low volume very large volume

    big data

    Big Data & Data Mining

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 7

    how to explore ?

    how to retriew ?

    Big Data & Data Mining

    pertinence of the result?

    how to analyse ?

    Data Mining exploration

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 8

    how to explore ?

    how to retriew ?

    Big Data & Data Mining

    pertinence of the result?

    how to analyse ?

    Data Mining analyse

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 9

    Data Mining Analytics strategies

    predictive analysis

    presciptive analysis

    descriptive analysis

    Data Mining

    collect large amounts of data

    try to give them meaning restropectivly

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 10

    Data mining Applications

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 11

    Data mining Domain of activity

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 12

    Data mining

    Data Mining provide a set of techniques to give meaning to available data

    How to mine knowledge from huge amount of data ?

    Prospection tools

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 13

    Data mining Principle : extraction of knowledge from large amounts of data

    build

    predictive models

    uncover

    original structures

    or patterns data

    data

    data data

    data

    data

    data find

    correlations

    between data

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 14

    Data Mining Specialisation area

    Data Mining - pattern recognition - data visualization

    - statistics

    - data base technology

    - expert systems

    - ...

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 15

    Data Mining Difficulties and challenges

    data data

    data data

    data data dat

    a

    dat

    a

    dat

    a

    compilation

    of raw data

    with different

    structures

    data formats dat

    a

    data variability of data over time

    data data data data

    continuous flow of data

    data data data

    data data

    data data data data data

    data data data data

    data data data

    data data

    data data data data data data data

    data data data

    data data

    data data data data data

    data data data

    data data data

    data data

    data data data data data

    data data data

    processing capabilities

    processing time data base technology energy

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 16

    What could Data Mining bring?

    example :

    - identification of a possible disease

    outbreak

    - prediction of epidemic before they spreed

    generate an alert if a precise event occurs

    ==> it could help prepare first-responders and other

    health professionnals

    watch in real time

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 17

    Why in Indonesia ?

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 18

    Why in Indonesia ?

    jokowi

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 19

    Why in Indonesia ?

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 20

    not any more a purely academic exercise

    could be a real challenge for healthcare

    and disease propagation

    detection

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 21

    terima kasih

    merci

    Gilbert PINOT

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 22

  • Climate change: Observation, Analysis and Health

    IICC Bogor, 27 31 October 2015 23

    usual data based systems

    high pertinence data

    DATA

    DATA

    DATA data data

    data data data

    data

    data data

    data data

    data data data

    data data

    data data

    data

    data

    data

    data data

    low pertinence data

    low volume very large volume

    big data

    Big Data & Data Mining

    Performance of

    new representation model

    are guaranted by the large

    data volume

Recommended

View more >