Prédire les valeurs d’une variable continue

Ricco Rakotomalala Ricco.Rakotomalala@univ-lyon2.fr

Équipe de recherche en Ingénierie des Connaissances Laboratoire ERIC

1

Tableau de données
Variables, caractères, attributs, Descripteurs, champs, etc.

Cigarette TAR (mg) NICOTINE (mg WEIGHT (g) CO (mg) Alpine 14.1 0.86 0.9853 Benson&Hedges 16 1.06 1.0938 CamelLights 8 0.67 0.928 Carlton 4.1 0.4 0.9462 Chesterfield 15 1.04 0.8885 GoldenLights 8.8 0.76 1.0267 Kent 12.4 0.95 0.9225 Kool 16.6 1.12 0.9372 L&M 14.9 1.02 0.8858 LarkLights 13.7 1.01 0.9643 Marlboro 15.1 0.9 0.9316 Merit 7.8 0.57 0.9705 MultiFilter 11.4 0.78 1.124 NewportLights 9 0.74 0.8517 Now 1 0.13 0.7851 OldGold 17 1.26 0.9186 PallMallLight 12.8 1.08 1.0395 Raleigh 15.8 0.96 0.9573 SalemUltra 4.5 0.42 0.9106 Tareyton 14.5 1.01 1.007 TrueLight 7.3 0.61 0.9806 ViceroyRichLight 8.6 0.69 0.9693 VirginiaSlims 15.2 1.02 0.9496 WinstonLights 12 0.82 1.1184

13.6 16.6 10.2 5.4 15 9 12.3 16.3 15.4 13 14.4 10 10.2 9.5 1.5 18.5 12.6 17.5 4.9 15.9 8.5 10.6 13.9 14.9

Individus, observations, objets, enregistrements, etc.

Équipe de recherche en Ingénierie des Connaissances Laboratoire ERIC

2

Statut des variables
C ig a r e tte A lp in e Benson& H edges C a m e lL ig h ts C a r lto n C h e s te r f ie ld G o ld e n L ig h ts Kent Kool L&M L a r k L ig h ts M a r lb o r o M e r it M u ltiF ilte r N e w p o r tL ig h ts N ow O ld G o ld P a llM a llL ig h t R a le ig h S a le m U ltr a T a r e y to n T r u e L ig h t V ic e r o y R ic h L ig h t V ir g in ia S lim s W in s to n L ig h ts T A R (m g ) N IC O T IN E ( m g E IG H T ( g ) C O ( m g ) W 1 4 .1 0 .8 6 0 .9 8 5 3 16 1 .0 6 1 .0 9 3 8 8 0 .6 7 0 .9 2 8 4 .1 0 .4 0 .9 4 6 2 15 1 .0 4 0 .8 8 8 5 8 .8 0 .7 6 1 .0 2 6 7 1 2 .4 0 .9 5 0 .9 2 2 5 1 6 .6 1 .1 2 0 .9 3 7 2 1 4 .9 1 .0 2 0 .8 8 5 8 1 3 .7 1 .0 1 0 .9 6 4 3 1 5 .1 0 .9 0 .9 3 1 6 7 .8 0 .5 7 0 .9 7 0 5 1 1 .4 0 .7 8 1 .1 2 4 9 0 .7 4 0 .8 5 1 7 1 0 .1 3 0 .7 8 5 1 17 1 .2 6 0