Data Science Lesson 1 Jean-Antoine Moreau

DDAATTAA SSCCIIEENNCCEE
Conference - Study : Jean-Antoine Moreau (Engineer - Lecturer)
© Jean-Antoine Moreau
copying and reproduction prohibited
Managing my copyright ADAGP.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
« The science of today is the technology of
tomorrow. »
Edward Teller
JAM 2 / 108
Contact http://www.jean-antoine-moreau.fr.nf

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
« There does not exist a category of science to
which one can give the name applied science.
There are science and the applications of
science, bound together as the fruit of the
tree which bears it. »
Louis Pasteur
JAM 3 / 108

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 4 / 108
Commerce and Research
are being transformed by
data-driven discovery and prediction.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
Skills required for data analytics at massive
levels :
– Scalable data management and off the cloud;
– Parallel algorithm;
– Statistical modeling.
JAM 5 / 108

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The basic techniques of Data Science
including both SQL and NoSql solutions for
massive data management.
JAM 6 / 108

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 7 / 108
• Examples :
– MapReduce;
– Algorithme for Data mining;
– Clustering;
– Association of rules.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 8 / 108
• Data Science Manipulation of the Scale :
– Database;
– Relational algebra;
– Parallel database;
– Parallel query progressing.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 9 / 108
• MapReduce;
• Hadoop;
• Relationship to database;
• Algorithms;
• Extensions languages;
• Key value stores and NoSQL
Tradeoffs of SQL and
NoSQL.
TToooollss

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 10 /
• Supervised :
– Rules;
– Trees;
– Forests;
– Nearest;
– Neighbor;
– Regression;
– Optimization.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 11 /
• Communication results :
– Visualization data products;
– Visual data analytic;
– Provenance (origin);
– Belonging;
– Ethics;
– Governance.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 12 /
TTooppiiccss
Graph Analytic :
– Structure;
– Transversals;
– Analytics;
– Page-Rank;
– Community detection;
– Recursive queries;
– Semantic Web;
– Quest.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
« Most of the fundamental ideas of
science are essentially simple, and may,
as a rule, be expressed in a language
comprehensible to everyone. »
JAM 13 /
Albert Einstein
A little break, to ask questions.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 14 /
Any e-commerce application is
a data-driven application.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• Learning systems;
• High-frequency / low frequency data processing
systems;
• Distributed systems.
JAM 15 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 16 /
• Aggregation theory;
• Data analysis;
• Data mining;
• Computational statistics;
• Statistical signal processing.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 17 /
• Kernel method;
• Probabilistic method.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 18 /
• Probabilistic Models
• Probability Theory
• Linear algebra
• New algorithm

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
The parallel processing of data
JAM 19 /
involves the use of
parallel processing platform.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 20 /
Service platform
for
parallel processing

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 21 /
A service platform will pool:
– Resources;
– Costs.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 22 /
• Using :
– Design NFA
• nondeterministic finite automata ;
– Design DFA
• deterministic finite automaton .

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
parallelism and processing performance
JAM 23 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
parallelism and processing performance
JAM 24 /
PParallel LLangage IIntegrated QQuery
(PPLLIINNQQ)

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 25 /
• Performance elapsed time;
• Scalability degree of parallelism :
– (number of tasks);
– (number of cores);
• Simplicity of the code (programming)
• Facility to maintain in operational condition;
• A method of using the memory;
• Memory Mapping.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 26 /
• Implementation method:
– class;
– objects.
• Method of using the processors:
– Physical;
– Logical;
– Virtual.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
Use the parallel processing to the databases
JAM 27 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 28 /
ppaarraalllleell aarrcchhiitteeccttuurree

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 29 /
• Parallel architecture;
• Specialized parallel machines.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 30 /
SShhaarreedd--mmeemmoorryy aarrcchhiitteeccttuurree
Process (1) Process (n)
interconnection
Memory (1)
Memory (n)
Disk (1) Disk (n)

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
Processor (1) Processor (n)
Memory (1) Memory (n)
JAM 31 /
Network
Disk 1 Disk (n)

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
Transfer large volumes of data in main memory.
JAM 32 /
Calculator (processor) saturation risk
Decentralized processing

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 33 /
Central computer
Backbone
Dorsal Computer
Machine Database
Data Base

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 34 /
• Only data used are transferred;
• The central processor:
– updates;
– research;
• Specialization of the computer;
• Use parallel machines;
– for data processing;
– for treatment of discs;
• Fault tolerance.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 35 /
Database Machine
• Control access and permissions;
• Semantic integrity control;
• Recovery Data Base;
• Atomicity Control of transactions;
• Restore of the database;
• Simultaneous access;
• Concurrent access;
• Recovery data base.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
manager Data
JAM 36 /
host system communication
interface
manager
Manager
associative
memory
query
optimizer
manager

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 37 /
Database Machine
• Interface database manager;
• Communication manager;
• Query optimizer manager;
• Memory manager association.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
Central IT Local network
JAM 38 /
Interconnexion
Interlink
System (1)
Interconnexion
Interlink
System (2)
Communication
Processor (1)
CP (2) CP (n)
Bus (1) – Network YNET
Bus (2) – Network YNET
Bus (n) – Network YNET
Processor Acces (1) PA (2) PA (n)
DDisk SStorage UUnit (1) DDisk SStorage UUnit (2) DSU (n)

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
 Ynet, an interconnect structure for a highly concurrent database
system which also requires high availability and real-time operation.
 The Ynet is the basis of a multiple-instruction, multiple-data (MIMD)
concurrent processing architecture which can connect up to 1024 high-performance
microprocessors to form a very-high performance
JAM 39 /
system.
 the Ynet results from the logic implemented in the processor interface.
 The processor interface is divided into three sections, each of which is
implemented by a finite-state machine.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
« In questions of science, the authority of a
thousand is not worth the humble reasoning
of a single individual. »
JAM 40 /
Galileo Galilei
A short break for you, to ask your questions.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 41 /
Database Machine
Processor
Processor Interface
Central Interface
Session control
Data conversion
SQL Parser (analyze – optimize)
Compiler
Dispatcher
YNET interface
PA YNET PA

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• Processes and access processors can execute
simultaneously on a single query;
• The uniform distribution of data on Disk Storage
Unit;
• Using a hash algorithm.
JAM 42 /
Specifics

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 43 /
Processing queries
Processing steps of a query

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The programmer examines (study) the query
rigorously and methodically;
• The programmer writes the query;
• The programmer tests its query;
• The programmer sends the query from the host
computer;
JAM 44 /
then ... more ...

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The Analyzer PProcessor IInterface (PI)
verifies that the query is semantically
correct and develop a plan for execution.
JAM 45 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The value of the primary index is transformed
into hashcode and is transmitted by the YNET
PI interface.
JAM 46 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• A message is sent by YNET, to all processors.
JAM 47 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The YNET interface of each processor can
accept or reject the message.
JAM 48 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• One processor acces * (PA) will access its disk
storage unit;
JAM 49 /
* Access Module Processor.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The processor access (PA) will build a
YNET message containing the tuple.
 A tuple is an ordered list of elements. In set theory, an n-tuple is a
sequence (or ordered list) of n elements, where n is a non-negative
integer.
JAM 50 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The interface processor that sent the request
receives the return message.
JAM 51 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The message is sent via the YNET / IP;
JAM 52 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• YNET sends the message to all processes;
JAM 53 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The YNET interface of each PA accepts the
message.
JAM 54 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The interface processor returns the message
to the computer according to the protocol
(TDP).
JAM 55 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The host computer is completely discharged
of the treatment response;
JAM 56 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
Host PI / CP
PA(1) PA(2) PA(n)
DSU (1) DSU (2) DSU (n)
JAM 57 /
YNET
PI processor Interface

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The compiler of the Processor Interface
generates the processing on multiple (PA);
JAM 58 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• Each PA will access the table on its DSU,
and copy the tuples that match the answer in
his workspace;
• A message is then transmitted over the
network YNET;
JAM 59 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
– of the PI origin collects the responses already merged
and sorted;
JAM 60 /
• The YNET interface

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The interface processor sends the contents
of the buffer to the computer TDP;
JAM 61 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
Processor Acces (1) Processor Acces (2) Processor Acces (n)
T(4) T(3) T(2)
DSU(1) DSU(2) DSU(n)
JAM 62 /
TThhee wwoorrkk ooff tthhee pprroocceessssoorr iinn ppaarraalllleell
T(9)
T(5)
Transaction

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 63 /
• Data protection :
– Fallback;
– Fallback clustering.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
« Science cannot solve the ultimate
mystery of nature. And that is
because, in the last analysis, we
ourselves are a part of the mystery that
we are trying to solve. »
JAM 64 /
Max Planck
A short break, to ask your questions.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 65 /
Primary Index
It is essential

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The primary index is chosen according to
the mode of use:
JAM 66 /
how decisions;
transactional mode;
frequency of updating;
frequency of application.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 67 /
PPrimary IIndex
• Sigle / UUnique (UPI);
• NNo UUnique (NUPI);

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 68 /
• The query on UPI uses a single AMP;
• The result returns a single tuple;
(Access Module Processor)

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 69 /
• Two kinds of access :
– Direct access to the primary key;
– Indirect access to the primary key, via an
intermediate file.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
1 n
JAM 70 /
Primary Index Value
Hash - Coding
AMP

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 71 /
Secondary Index
• The RDBC build for each :
(Recursive Density Based Clustering - datamining algorithm)
(Root Data Base Connectivity).
– Unique (single) Secondary Index;
– Non – Unique Secondary Index;
a sub-table.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 72 /
Sub – Table
Sub-table contains all secondary indexes
encoded in a hash algorithme;

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 73 /
Sub – Table
Each sub-index of this table is an identifier;
the hash code value of the primary index,
retrieves the tuple to select the DSU.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 74 /
Tuple
Size
Secondary index
identifier
Secondary Index
Value
Primary Index
Identifier
Tuple
Size

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 75 /
• The value of the secondary index is
transformed by hashing into the AMP.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 76 /
Secondary Index Value
USI
Hash-coding
AMP

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 77 /
• Each sub-table has a column identifier;

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• each value of this sub-table is one and only one tuple.
JAM 78 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 79 /
Secondary index
not unique - single
structure of storage

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
Secondary Index
No unique (NUSI) Hash - Coding
AMP (1) AMP (2) AMP (n)
JAM 80 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 81 /
Query Optimization
SQL Query
parser
resolve
optimizer
Structured data
Explain / Text generator
dispacher
statistics

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 82 /
• example commands:
– collect statistic;
– explain.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
The COLLECT STATISTICS command
JAM 83 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The optimizer uses the statistics collected
on the tables in search of better access to
the data.
JAM 84 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 85 /
• Number tuple for each table;
• Unique and no-unique index:
– Primary;
– Secondary;
• The number of the attribute values;
• The maximal attribute values for each the tables;
• The number of occurrences not duplicated in each table;
• The collection of statistics is made in order.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 87 /
• This command provides information on:
– The nature of the requested tables to process the request
– The indexes used :
• Primary;
• Secondary;
– The various stages of the request processing ;
– Treatments of joints;
– The contents of files created to stock the intermediate results;
– The number of returned tuples at each stage of processing;
– The execution of the various stages of the application time;
– The types of joints;
– The total time taken by the RDBC (datamining process).

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 88 /
• Syntax example :
EXPLAIN COUNT ( ….)
FROM .. , .. , ...,
Where …

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The minutes (report) of EXPLAIN,
– the time set by the system to scan the tables and
JAM 89 /
selected information.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
It is important to make an algebraic tree.
JAM 90 /
• For each query:

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 91 /
Query processing on RDBC
– Recursive Density Based Clustering
Query Single table

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• Query on UPI (unique primary index);
• A single AMP is accessed, and will look in
the DSU the tuple corresponding to the
query.
JAM 92 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
– One AMP is indexed and accessed the duplicate
tuples are placed in the same GPA.
– The GPA will look in its DSU tuple or tuples
satisfy the query.
– GPA: Grade Point Average.
JAM 93 /
• Query on NUPI:

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 94 /
• Query on :
– Unique Secondary Index;
– Non Unique Secondary Index;
– Request on non-indexed attributes.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
« Science is organized knowledge.
Wisdom is organized life. »
JAM 95 /
Immanuel Kant
A little break, to ask your questions

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 96 /
Processing requests
Processing Query
RRDDBBCC
Recursive Density Based Clustering

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 97 /
multi-task application Request
multi-task Query
• The RDBC uses five types of joins :
– NESTED JOIN;
– MERGE JOIN;
– PRODUCT JOIN;
– EXCLUSION JOIN;
– ROWID JOIN.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
• The wwhheerree condition specifies a constant
value for a single (Primary or Secondary)
index.
JAM 98 /

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
join two unique primary index (UPI)
JAM 99 /
Select … from … , … where … ;

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
Select … from table (i), table (i+1) where *;
JAM 100 /
MMEERRGGEE JJOOIINN
object-oriented programming :
Select … from object (i), object (i+1) where *;

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
Select … from object (a) (i), object (a:b) (i+1), object (a:x) (n+1);
where * ;
with *;
JAM 101 /
PPRROODDUUCCTT JJOOIINN
Select … from table (i), table (i+1), table (n+1);
where * ;
with *;
object-oriented programming :

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 102 /
NNEESSTTEEDD JJOOIINN
Select … , * from *, attribute
where x=y and y= (variable integer);

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 103 /
CCAARRTTEESSIIAANN PPRROODDUUCCTT
Select … , …. from … , …
a PRODUCT JOIN without restriction.

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 104 /
EEXXCCLLUUSSIIOONN JJOOIINN
Select …,…
from …,…
where x=y and (select … from …);

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
JAM 105 /
ROWID JOIN
select … from X,Y
where X.a = constant
And X.b=Y.c
with
a : primary index
b
c secondary index

© Jean-Antoine
Moreau
copying and
reproduction
prohibited
Managing my
copyright ADAGP.
end of the internet diffusion of this lesson
JAM 106 /

Data Science Lesson 1 Jean-Antoine Moreau

More Related Content

Viewers also liked (15)

Similar to Data Science Lesson 1 Jean-Antoine Moreau (20)

More from Jean-Antoine Moreau (20)

Recently uploaded (20)

Data Science Lesson 1 Jean-Antoine Moreau