ADS Capstone Chronicles Revised

16

Td oh i ns oi st dd oo nmei ns ao t tehtaht ef ema toudreels, lwe ai tdhi nl agr tgoe ra smc aol er es swt ai lbl lbe ea nbdu ei lfdf ei ncgt i vae rreeggrreessssiioonn mmooddeell. Gwi vi tehn twh ee gr aotael , socf a lpi nr eg di isc ct irni tgi c sa tl ot mo amc ahk ec asnucr ee r f ema tourrteasl i toyf lma rogdeerl s ci sa l ems odroe n sotta db loem ionvaet rea ol l u, r amn do d ef ul , t ut hr ee potential regularization does not dF ii gs pu rr oe p o4r.t5i o nbaet leol yw i md epma cotn sct rear tt ea si n hfoewa t u rtehse. St ht ae ny dhaarvdeS ac pa lperro xpiamc ka at egley tshc ae l seas mael l mf eeaatnu r( ez es r soo) aFnigdusrtea4n.d5ard deviation. Data Normalization and Standardization Visualized on a 2D plot (Images Below from Cosgun, 2023) Wt h ah te nt h sec Sa tl ianngd oa ur dr Sf ceaalteurr et rsa, ni ts fi os rcmr ue cr i ai sl tf iot no on tl ye oi sn ft iht et efde,a tiut r ei ss f ot hr etnh e ut sr ae idn i nt og dt ar at anssef ot . rOmn c tehiet fTehaet uSrteasn df oarr dbSoct ahl et hr et rtarna si nf oi nr mg ae nr di s tne es tv edra tf iats oe nt . twhhei cthe s t wdoaut al ds e t atrot i f ai cvi oa li ldy a ni my pdraotvae l e amkoadgeel, ps cearlfionrgm, tahnec et r. aFionlilnogwfienagt ut hr ee tdraatians- et et si ts s2p0l i0t , 4a n6 d9 rdoawt ass eat ni sd6 65,88 2c3orl uomwns sa nadn5d8 tchoel u tme snts . f e a t u r e

4.4.3

Preprocessing

for

Modeling

Approach 2 4.4.3.1 Feature Selection Tu shee dgi mo ael nosfi ot nh ai sl i tpyr reepdr uo cc teisosni nt go ma papkreotahcehs hi se et or aHmo wo ue nv te ro, f S PDrOi nHc icpoal ul m Cn os me apsoi ne re nt ot s w oAr nk awl yi st hi s. (vPa Cl uAe)s , dsooe swne omt uwsot rfki r sotn haa nddal tea st he to swe .i tOh f nt uh lel 23 7. 776, 8%9 4h at ov tea nl oobds ae trav af ot iro na sn yw Se Dh Oa vHe ci no louumr ndsa toar, Ct hDoCs eH ecaol ut hn tMi eesa swu raes s cpor el us me nnt s bi ne c tahues esdt oa mt a af cohr ct wa noc oe rt hme ro rdtaatl ai tsyedt sa. t Wa s ee to, bp ut tt on odtrionpeti ht heesre orfotwh es awsi tthh ear en waillml boes tv eernyt il ri tetllye- wmei scsai nngd of ewa tiut hr et hseemt . T 35 h 2 e r c e o s lu u m lti n n s g . feature set has 267,444 rows and Fo bo lsleorwv ai nt igo ntsh, e wree mi novveasl t i og af t et dh o swe h i3c .h7 6S%D OoHf ct hool us me nc so l hu amvne s r ea mr ea irnaicneg- rne ul al tl evda lcuoelsu. mMnasn y( f oorf et hxaatmips l eA, spi aenr c)e. nOtua gr es ot of mp oa cphu l ac at inocne ri n ma ocrotua nl i tt yy dc oamt afsoertt aabl rleyaddryo pc at ph teus reecso rl ua mc ends .aTt ah, esfoe ww oe t chaenr cr eo gl ua mr dni ns gwhi tohwml oa nn yg nt huel l sc oamr em: (uat )e f wo uorr kc oe rl us mi nn as ctroaunnstpyortmatuiosnt tias keandto (bw)orktwoon colpuumblnics rheoguasredhi on lgd sm. Ge di vi ea nn hi no cwo mm ea noyf ngur lal sn da rpea ri ne nt th- el es de towt hoe cr owl ueml l -np oc aptuelgaot er ide cs oalnudmtnhsa wt tiht he irne tahree SmDaOnHy di nactoams eet ar engda rhdoiwn g mgaennye r wa l omr keedr isa nd oh onuo st ehhaovl de a th c e ce se ss to cars, we are comfortable dropping columns as well. After the

107