Este proceso generará un código de analítica predictiva el cual podrá ser ejecutado en Tableu o Power BI. Seleccione los datos de entrada.
Variable Objetivo: DEPOSIT Valores: yes (naranja) no (azul)
Se corrieron tres modelos con diferentes sets de datos. El primer modelo es considerando todas las variables significativas, el segundo modelo elimina la más significativa, el tercer modelo elimina las dos variables más significativas.
Data distribution: no: 52.68% (58796 instances) yes: 47.32% (52824 instances)
Predicted distribution: no: 50.49% (56360 instances) yes: 49.51% (55260 instances)
Field importance: 1. duration: 32.96% 2. month: 24.19%
Data distribution: no: 52.48% (58583 instances) yes: 47.52% (53037 instances)
Predicted distribution: no: 55.51% (61956 instances) yes: 44.49% (49664 instances)
Field importance: 1. month: 31.73% 2. age: 12.07% 3. balance: 10.37% 4. day: 10.23% 5. contact: 10.18%
Data distribution: no: 52.55% (58659 instances) yes: 47.45% (52961 instances)
Predicted distribution: no: 54.44% (60763 instances) yes: 45.56% (50857 instances)
Field importance: 1. age: 18.29% 2. balance: 16.72% 3. day: 14.51% 4. campaign: 11.05%
Data distribution: no: 52.68% (58796 instances) yes: 47.32% (52824 instances)
Predicted distribution: no: 50.49% (56360 instances) yes: 49.51% (55260 instances)
Data distribution: no: 52.48% (58583 instances) yes: 47.52% (53037 instances)
Predicted distribution: no: 55.51% (61956 instances) yes: 44.49% (49664 instances)
Data distribution: no: 52.55% (58659 instances) yes: 47.45% (52961 instances)
Predicted distribution: no: 54.44% (60763 instances) yes: 45.56% (50857 instances)