\ Wit Makiot Leis) a
oe
re =——2
» —f, 2 ~ 4
|) eee ress ee a |
+ Pale |
- AT 3 / a
_—|_ > ea eng
> a Ask Same Roe
human behaviour)
3 aipeoak of Tu & OL:
TAL?
> Shands pow aa Lewin
sy
|
= of AT.
She nc egy tl esichey ss
> Stands
Mechine —Leanning “Mochanis
> male He tabpolation Of
— mull = pyen
m
(oP
5 “Pf Leeroy —
TYP
= 4
ttt ' Ss —
1 ASS ety
-_
_
a
Mo
a
—a7
=
om
_
tL
|
|
|
laieevrril o8
Date a }
re aa | i
i
7 Pepplications —of Machina = so Redan pn ||
“Thou ast map — alcatel Atos
1 Goes lo THs: Fasten Rowe Seldion |
| Shae -athich, eae Beever a
V0vTT777
ee - a:
> Bogpite of thr usual tralics You ou _—
~~
~~ onthe fastest route — ——_~_____
— > _Evesyone _usho 35 using the | | aes
=
=~ rip 38 conbi bubiny ah, ei einane
— accurate: “ = at
aad oe peg Regen —
= > Feacsbook Su iL you wort toto.
a the person Sin the pt __~
ia] = > Uns —-focebook's Sau detection -and
= swocognition -cigorithms ah
= “Ba on tid Siqjed “Borpbact
E 3 UBER: ML at Uber —_
hap 3 Burso: nalizad Application a nS
oe | ES mated “Tmt _of Arrival —____
CT [4 Pda De commendation oa
kag > Recommands ada based -on qo Exar —
i _ |
hskg : J _|
| oe 45s in gg enerabing —sucoran endation “by ny
—_ masten chef on “the usorld- oc ne
— ob An mmazan's_revenur iS
~ by HE -cuttormmundation -345tem- J
Mel
—Sraorkphone — utith focal Fec99 oiling —_————
Coe of the -faco -clleckorJ- machine ——
ee calaorithm
— = Pokdefay Fai chips:
yD wy
— =“ Priverless Cars —_——
A “Teslels Als deen by ‘Nvidia's Hi
—____ focusin: —Taainly on aupervised Loni
ak Dtrowdsedeas A 4rom—al)_- of ats vehided
———and_its drivers - internal & external
enor —
int
__dimlar - to foun e
=A 157. of ureng Selects
—__based_on_-Nebflir's Fecommindalion
tng ml
movieX
he &obalic Chot -
can -uQlicates AiSh Cooks
> _ ah necorda it's
ashes Cooking hiskany
—— en fundion ee 4
COCO)
r
) Date 7
Page
777
4. Gaga ranglate: ae
|
S |
a3 aos e 18 No mone -q_barier!
> fer, Fast eo =
__{rang-\ation-
Real - biew translation —Comerq
TITTTTT
Al
kKUKR,
> KuKA iS a Crorman_monufacturer of
—|______indlustrial_robots and Solutions - for “Facto, —
Automaton: _ pit a
FIGTTIT
L
en a al enabls a machine to_Qulomabical
es = an ioyprave Rorformana from _ePeriencss,
SSuedlid thing s__ctithowt Being explicitly __
vIT
SDros rammed
38 Stakes et nh ete ea
}
» dhe term machine torn 100s $irst_intrduced_
yy Drhur Samad in lay
+ utith the ful of historical data 1 -ushidh is
Kroon as training clata y_ mf rithms build a
rathematical ol that__helps~_in_make
C predictions -ar_de cisions sith out beg
CTT TT
—Eaplih st rammed a
+ TL brid and_- Stabistics sAggethan fon
ttr
AT
trucking edit __rnodels i monk ho Qroduckion of data —_—™
va Cn hs (ich cou iil) — em
foc dot bw -<
lee oi ous, nae
eo
| + & _ hidden ‘ peed and_ex ——
i
Fealure of Mi: Los
-
TL uses data to detec voriaus Sabena
rere ~
ee ee are {tain from SPask data ond taspinve — _
automatical ont
» i ig - driven -technol: oy: aa
ML Ss: ammuth Similar +o -cl as
jt -also deals - usith tho ing can
the claita
POET S |
=
rm — 1
Need_of MI =
> The ned of Th as inowasipg dey by
son behind the need 35 that
Tw
+ The _vuc
3448 capable of doing that ose too
lex _for_a V4o Sle mant_clredh
‘ As ¢ Se teh et
ec not Access Ce
yo how o L4to
make it easy.
+ So explo the fata -
. Constrack tha_modal
Now
+ 8c redict the ouquired o/P automaticaly
A machina hog the abilily to ‘leam -if ik can
improve - its Specknedoance toa —-Sintng mmace choke —
Mechanism_ol ML
> A ML. Suslem barns trom historical dala. _
Builds the = predlishise modes Qc _cshonerer
ik mactives Mud cal ba eR adits Uae ey) for jt.
dht_auuaw ot i ded_ubpat
Hon “bos. Dai. ok _ data ax the je amount
dota buts to build a bethen moll ahh
TEV IIFITITTITT ITT
adil thi _O}f) rare ae
yyy
u
j | > inttead_of —ubitias pga a oc tees on
e ondhms, machine ——
baila tha lpg a5 Ger_Ae cat” and —Psedi “as
aca
3 Tl hos _dhorged unto ‘of thinking about — prom
| ua SQrabla m - & ns
| en
1 Sek Diagram = my
: _ os
i araining [> 1 ¥ eT a | =>
S y) TL Building legica
[aS tis [meds : a8 =
| { Learn trom ; : —f —
| T dora 7 Reo dala ;
; = ate
| Marhine earning life Gide gy
[ | a S
Wee oe
= soe
ia 4. hemos} miGortan! thing an the Confplabe
= Coprocess ‘96 “An understand Ae prob od is
toa Knox the Shaigost ot ee sla .
ben 2. Therefore, - be bey Ute nai)
—_ to _undarstand tar rc na baer ra
-_ do Mex ondarsianding ¢
= Codreblem -
wes | Fn _the -Com@Plole lite dle Grows, to— Solve a
lum ,--.0e _Crtocde a Phachine _ Hints im
Catted “oc oC at nodal 36 a
Skeet “Bul to train te
aan
(ce) Rag dharts ay Colle ding
ae
P
| I en NO
+ The goal _of “Ais -§ F skp. as aden a be
| ——-Sblaite_-all_data - gulated SpProblems..
| Bala Con she collched - trom_difken!—
| —-Soures_- Such _as_- is, -dolabase,—anbernst —
b> © tmobile - doniicos.
Phu quo sand gus “of the Colaced — oormy
~data —usitt | iitermint
thiciengs of thy — =,
of. oe
>the rere sill be the -dalo, “Hne_emare ——
;—Aclurake will “be the “redicion-—_____—
— prs She indades Lhe - below -hasks
> ddenk} Varies Sources
=) Colle
edlaty
= Inked nie qrake div data -cblcinad trom -Ailhrent —=
ylang tho -above task fe ga cates
Called -a3 a_clataseh it ~catill foe
Data © PrfParation
=
und jn» further Steps. my
: —
‘aay
Alter Sais the data, en ned 40. Qrepare
Sk Sorat = Steps
- amr
Data Figecctalen' tao Step whore se Qed
“Le far sila jndo -a aS fl
here, ote clfploy the _model in tho sual -usortd
_ Aygtem =
alr he above refered model _ is Groduciog —
accurate susult as et cur_staguirement_esith—
TOUT.
atteplable Specd + Athan_ue_-oeey _ ppodel in —___|
he sah t —+Qandas callous — us = “anal a sq_data_ al mg
Conclusions fused 90 os -a, theories _
Sanda __can__clean_- “dala Ses, Beale
__ them —_staaclable_-and all fbs etn
oe
z
EF Revork dota is ey id ‘nfl tin data
Dooney a
Hh
|
in
Bula Scena: 25a “branth- “of CS, ushurw ite —
hed hoa +40 _Stosu-, use bc ana yes_catet—
| for desis Jnfocrnation- $com it~ ——
_ailzo_ able bg dale “roi hat on
ins _*\res faluos., §ike
Called cleaning —
—) Spandas —a110—
_ | eb seelevantt sc _C
eos .er Null _ aval
| Ane — Y dala —
|
H
Bape ig ——— 7
ie
~_puepork. Sya
eee after
ne ur "4 ee qd
=]
Page sy
a li aL". uw inh
ES a] books Glo’ Sule —— =
Rms ‘guthoce!: eee rtp iy
E "Chale “Bhgpt 1——__ ey
E Deaton ik Libra /—_——_— >
rc aK mae) —
L he
if Jandas 5$ -usuc else —— 5
K alias.
[— ond then eto Se Gah
| pa Pandas =
C altribale - ss
[-— Grint Pde version J =
[- | Benies: ad =
[ 2 inane nieg 25. Dia a Column ing =
i “lbs a ono dimenSenal arrest ing —— nila
L data of a bfpe: _ g a g |
E Ea: init Pandas 38 —s
: 01,9,33 i
I ‘ter Pd. Se. =
: ~ pani ten) on —
f %p—°9 5 —
Fd [ zy ae a
re
eee
a m1 ee ie afoul
Ss
| Labels: by
— __ inspect (pandas as “hd
as 2090
| —____ den’) = Spd Series)
VITTTIT
: ts Co thc te)
opps i
|
17
__Gastom habele'
——__ ieipoct “Pandas os
fs £7,927)
|} ——___var = “pds Series Ca, dndye'= [een tyt t2"])
C peink (vo0)
Cqrink (vacl 9)
Op:
x—1 _
TIT7T
yi]
|
ze
ITTT
J
ht es
Converting “Dickenery as Suxies,
apo Pandas _as pd
P| calories = $day a0. “aye” : 260, "doya?s 336)
bey | yaar _= Pde Seniest calorie SN
o Spitinhvore)\ stew = fe ys ee —4
foes 7 fie pa —_
— “DetoFroms —_—
a Dela -Sels in Pandas _oout_tusudy owt - dimensienal
ables , called “data Frames
VTTTT
> Series a8 ke a Coloumn 2 “dala from |
48__+the shale - table ‘
a
TT
[3 A GQandes alaFrame 15 _a_2 dimengional dala
| ___ghewkua > Vilke Qed array» 9Fa Aable_usith 108g 4
columns.
ee ee | lala =~
* Calories :: {420,380,340 :
| “duration”: [$0,40,4S — ——______
3
; df = Qa Balahrame¢data)
—Oprina
Calones __claration
——— 420 bo
‘ B80. 40
as 2 330 4S
Locate Row
CQandas__use loc alvibule to cuhun one or
mmore__ Specified tows)
Ext
CPrink€ d!-locfo)) # cefer Lo the Fo ondox o
Of: Calorie: 420
duratien 50 4
Idk telus Candas Ses ea
Txt: _
Print df-toc{ fo.133)
orp: Calories uration
> go So
, 2 Bow | 380 do ‘
_ushen ute au usin 114 braduk
the _susull as a Tencks “DolaFiomes |
. i
TT fs,
| Namd Indo xes:
——uatith Abe — index apy inend You can snarne
7 ~d ousn indax es
Ux: defPork Spancas as “pd
_____ data_= 3
‘ Calories” | 1420, 380,390,
+ duration": | 50,40:453
VVVITT17777
x —|}— At = ok DataFramal data, inder = Ee day™ “da aft
Ss = “ dgy3")
~~ —— rink dh)
~ Om: + ___ Calories duration
St ot 420 So
~S = bh 3RO__ 40
F 3 290 _ 45
“Locate Named Jecloxes
__ CArink Cds loc £* doy" 1)
op. calories _380_
_ duration 40
Load Files -Into_a CQataFrame “Ucsy)
a your data Sets_-couw Stored ana fl
aS can ‘loacl_-+ Inara. into _a SHataFrame
TIT
He
es FPardos 05 FC _
aad ds read mnie data-csy')’
coll “gach roiel ai ;
TIT
|
i )
7
Pi
))
[Seiad COX Files Comma Zeparaked files)
ae ——w
y)
CSV ( Comma _Zeporated xValurs) Silas contain
Drain, Yen tnd hal Keno ferme thal =
WD).
Oxy r —
ead cov C'data cay) __—
At = dere
| CPrnk (at to Shep) _—
[3 eae Sig in Spi ha rae Deal
i Mor moag: 7 ee =
>The _numieen_of ros -retuend is. deh ey
. in Sehndos_cPlion - Sethigns fame
- Fu idport: Sandas_as 9
[ Brink Gol cpions: dis Play. mox.roas) —_T*
b | OM? bo _ ==
i | Thewase the maxima i tod
| wie otis See ee Sa
| |
Bic infpork andes asd emi
Cod: eplions -clisplay +max_rows = 9499
- df = Dd: read csv C* data. cay’)
—
rind —e
“Prac JSON C aaSeral Obieck Nol,
j Oli \Oiti Kan)
a
7 7etts coat a
: “Bie a Otten Stored, or_ewtracted din Son
+ QS0N_is plaintext, but has othe Scrmal
—j— of _an_objed. As tell Kenyan ine
—| _uterid of Croscamming.
= ngeecmigg ;
17.
load a “python distiongy— into. asd akeFrame: i
“iniort Cpondas_ a8. Gd_—___ |
data = 2 gui wrt Ds
“Ducted § 4
Mot; Go, sis. _
aie Gouin etree tron 1
na" bo pitti 108
Ml
TTT
\ | ||
VIVT7
J
a “gh 7, sis
WW
t
_df = Qa Bata Frama data) oe es
SpriakCat) e seat
VENTS
Road) _ sa gfe wa.
i 4 ©Qandas cH wd eet i
“igi fecal json lata: sen)
prank dhdo.- String) god BO be —
_
VET Mie a the dala i
—O theadk boadQ =
|———>_sutlurns_ oe hvoders anda Specified
cnumber_of rows, Starting from the top
Print é df beaden) a en —
~ if the numbes of rots “is_nol_Specfied,. —
Lhe heacle) rruthod tilt setarn the 45 routs. — emmy
@ sail. a,
2 owlunns the beadew and specified Quraber—
{+ “See Starting from the bottom»
ni df. feild ne —_
| @ tale “AlbeuL Lhe’ Dala —
The Datahames, -@bjed_has a_tmethod Called
-— infos that_ gives you_mraru—shformation Ss
—abeut_datosel —
GQrinkCdl- into) =
oy
WW
j= _aiso tells us hoa _ma -oul
jj ———Naluts_ thew ano Rowsent an each Column» —
ata Cleanice *
CData cleanin
—mgang fi
jeu “Adley Ataing— ‘Bad_data in
at data Could be: E =
“eo ae Foca
+ _ irony
— Baie ‘Cakes
=
—
nea
=
=
Sa —
Ey tall i
oe!
et
et | ahply Colls
>Frsply Cells can =iherl qe cjoua wera.
weesull when yeu anal ala |
Wools fo Boal with? ob) C 4
) Wotmove” Hows _|
> one Wo to «coal utith me cally 35 to
oumove J stout that contain e Sy cll
Lx:
gator! Sqandas as Gel 4
dl © Pde vead -CS¥C* dala: csv’): ad
n0o-dt = Af deep) ts
Spork Cnod dH -ta-strieg 0) a
sy Py default, the dréprad)_methed clung a nea -
dation, Ar esi nal chapge the eciginal SDoteFrams
a ye uoant to change £ oZyiea Doe,
= the drlace =“True a ae 1
en ca zs
cle “dh reacd_Cov (\ dalascsy)
Ale dvopnaC inlaw Pnolher wo af dling obith enn calls isto
ongerk a Nolug -onstead 4
his wet ou do -nol_have to -dolebe entice __|
Touts,
PUREE TIISSTIIITI TIT TI TTT I IIIS | + ndas_aspl_——-~
| | esc read _5y ('dalaec st)
| “Abo Silt sin Naina Tus) ———
Fela Gals for Spealied “Coluens ———
Colurany =
ie “Spat ese me: —
imo
Column nai _
ea
0: Jost “Pani flac.
df = Pa. owe
ALL Calories") -fillna( 20, ox place =
[Axploa bing Mion. Median _or Mole
» A Comman wo do. eu
— calla the mvar, median - or -mede
radian()_and soerlet) mal
lorieg” fla nal x, “ies = "Inn 9
the averse volus “MoclianQ) —- ____— Apne at
i— iar FPandas_as Pd Sit ee one ieee
eos dit = Gul stad_csut ‘dalavcsv)
X= _Alf* calories") «modianty
df LY calories" J. -fillnal X, doplace
Tru)
Median + the value on o_Staa amis ale jou hove
— ___can_-make ifficult , or -even.inipsshl fod ci data.
ic Jha 28 a tuto si:
i t- Cony rerk. Indo. a Correct Forma}. |
pecs + -Seppose ate — hove a fle_in- -ushich Bate _formal
a
— asi Libsiton' _|
fee — Sl “ah Chinvsth all “Calls inthe! Date Column — -
i gato_dates -
i
Lo
e Tithed ——
a | SPordos _hns ao “to dole)
Af = Qa: ead cou dob esy')—
ALL Pate <@d- to dote nol ia ——
Prick l Af. Ao < Strigye)) — eo
L CRanasing” Bas = SS -
\ > the exarkpla -aboie “gave US 5a Ba ohh — ithth
i—— ye handled as“ Nall valu -_—
[ak cra Can amore tho 10d By ust I ca big PES
iT
—s
OO '
|
| ex:
t i a: dapat Sabie = {Baten
| [Fang Wiens Bal bala —____——
data” does nat have 40 be *
> Wi
oct ting nn i con just bea
[_Someane a _agstead_-of “1
Thome ase wat
a
CH Afelock1 Duration!) <45
Tor. §mall_-dala els - ?
. |
eee 0 [A j
3 Io vwplae wren dota for dayger -clata -Sels you
Con cate - Son rus €.4. Set ~ Some boundastes
— for Sogal values, & oplace Pratt ot bondi of cee data is to
| __amove i rows _ nat orleans mops data.
ex: a -
- Sor x in dt: inden:
a a loc Lx, "Burak
a ___ dh-drop fx, seplare =’
Be moving a Ha licales tps i = _
biotig soe Sate a
We discover. duplica ates, ule can use the duPlicalek:
ropthod. _
2 U_ayolurns_oBaleon_voluss for oth ro
print dds deplicated) a
sulurns trur far cuiplicale tox else falge
lovin. AS
26) nove
___et_ dds deo apices inp Te Le
(irlat = Tas) eaill_sumove delicate mail |
( | ts
ablframe
abe. | Goad Corre lalian:
Folin alogys_ has 6a Spe
a—Cortelabions
-—Fidsg Telkcntp ee
ee
———— =
dhe Psat 6 of the -Comt) mothed -i5 is a 4oble
-
—aiith alot of numbers_-+that otipreSenk -how) wal
—tht_oulabisnshp is bla_2 Columns. .
4-3 Nember varies fiom lio)
eee q
|" evdecl Comelahan =
——whan_- Sams Column file -“Durahion" & Duration =
— Jak the number 4-dc0000.uShich makes Senseah =
feck gutokenship asith del —s
—“Buraken® & “ Colaries" i a “os 982721 -¢ tain
=
es
wthicn 5S a “a Jjoad__Correlation & «oo cana
| edia Ahat 4 (et ter out aha -tmore
— Calorie s_ gjou-burn -Othor uy .areund: if _
—>———
Burnad_-o. lot_of pind fob -bad_
caer *f ook uk ames oo meat sh
tach Correlation: _ 7 a
“Duration” ke *Magpilse* acta O-ooadg - -Coyehen
Bese y bad__ ce _ Con Quedd Bucdien _ MYA PY
tht, rn
I Ploling
| 4 Qandas
| > We con
ed
ai =9d
dL platO
Goa lal
cater Plot
= - Scatter
cette
Jo viSualizo the diq.grom on_the
oryPork Ban
oriport rmatplobb “pt
+ read csv
2 Specly thot of Wank a Sco
Kind argument
Kind =\ Scatter’ _
atts Bu
es glotO mmothed
use Takplotlib Mibra
das as “Ad
0)
@lok_noeds an -%- br a_y-anis
““Bauolioe! rye * Calories! >
QPenvised Lecrrisg_ =
sp They. of = Rip eubieh. pean Lane — x tata
yO] foal re _dlocka__and__on—
data. machines pot
———ffoith dhe _Caserecf rut pet
lerbelledl dala maars sore inp 1 abenely. toy
Wa)
The frmining gato _pasvided tothe ei
(Sf the _suptnaasonthal-—Jeache? ~The _raackiner do pie ean
Wide the outpul> —canmead dba cpplizs_Jhe—tarne 7
asa otudenf- — Jeauns tn the, Superwslan_%f— meme
Fan x te a
Sb bh a pce — idirg. Input alate eels
a Pte _model_» “Thea
Caxneck output daty a dhe v/ cS
of- a Bey erase Leos je ak euthea —f ie as ihe
Leoith Oe, cutp Se =
: =
wo. Pete? of sipevised deadtning”- =
|=> Th esa leo rodelr ane _tsainge) usin —
| Jabelted daniel whe the tn Jel j oe —
each Jype af cada _. Once fs
Compeleted . the, model as te E
deka CA subse af the dh
puedltts the osbpat
dent 7
Bo reccwernsry Tag |
|
re el |
1
| deeps Involved in Supervised Jecrining? Sel
>| Heck detemine the Jypa af rmonning clade 6
Collect / péuthen the, labelled —teainings date ___ |
plik dhe Jreaining- clatarek into 4 airing. dlalaseh , Jeuk
| deutasel, ancl valiclation -dataref
Delewmine the inpuk features af the, training deuuets
_Jtwhich_- ghowlal Howe enoygh -knovalenlye do Shot: the model
foun aceuately peice “tha _ovlpudh—-——
2 |Determine the Auilable algari the fax the model, such —
as_cuppoxt> vecot _machind decision tree ehe+ ____—
@xecute the. algas}thm onthe. —teaining-dalasek gone I
toc need validation seh az the central panametes
twhich cure the, subse af — Jeaining datas aie te Da
(valuate the, acewtacy af the_modill by powell ing the dest
ge. TY the mode —prteclichs th couse? Outpub which
means oun model {1 aacewsae- _
ITITTID,
I Types of _Soneuvized Machine —leaxuitg” Algosithos =
FIITTTIIT)
HW
f
y Regecession palm
- =
[A _seqrcession pel
wa_recl value . Sv!
lw
y
ed ip there it 1
andthe _owtpal> -yeuciable - EF ts |,
such _a 5 Weather
of- _enoHinuans aeariable. ug
planet ‘trenely @k+ Bela we Some poplar H fe
ct
alggnibhons__ohlahctine— ancl supervised - |
il ~ oe _ a ne |
secastitees —
ion.
PUTT
ys Y
ip
= Ts
=e
Linean se giessian
99)
“Keacession Trees
Noi fioeax?‘Repevession
di
)
nec Recession
ae
Q._ Cllr esidscasi
i
ble geben Yr ulpu sete FE
_ ge ss h os Red * ort blue’ wor disease —
and 4 no di
| Castilian algoritms nates Sh
. Caan aed Ceska hich sant — sama
Pees ere as ages No.l lets — py
i
pr fe
Span_filbesuns
| -Rondom = forest
Decision Loree
c logistic 20 que on
pps ett c_madyn.
Advonluses
pacman
D> yeith dh help of Sipecwised Homing the medal
Can edict the _cutipet tha basis Of Dede
ence
(CORED)
| =. Biperstisel Learning _crevlel
—* To Sdperwised cLoasuning , nto Can - have -an -exack ot
[ies abasic ec Se i"
a a ae ale Vagus
teen Heal- World Problems - Such_as Freud -detedien, Sam
VIVITITITIZ
|
—
| Visadvanlages:
1 Bpeciced boon
—| __training data gets a
| > Traini -sogauired -kids - of Coputakien tims _—
I-A ‘In __Attperwised — hrarni:
Filtering ofc mss uly a ee
ning ~anedela cv" Suilable fon Fraraing.
——stho_-Corsfplox_-tasks " __ —
_— | > S@enuised hoo snip “Cannot CRowdich the Cometh
—_eulpat if tho ‘hes -clata is _diHownt from -tho
1
about the -classes of obied
cato_-Reed enough ‘kool dye. 7
ORR ERE
ex: (cla ification) Be -
= To ofp
Sap Cat =» brain => _Cepy_or_aal)
“den Proteu S Painlal Bain me ainfall in Gna
in cm
nit
|
-
~-
ot
qe