bank-marketing-strategy-analysis

Final project of Categorical Data Analysis 2020 fall.

Goal

  1. Forecast the success rate of the telemarketing calls, given information of target customers and the campaign records.

  2. Customer profiling for a effective marketing strategy.

Data Preparation

Data explore

Discrete variables

2-1

  • Personal Information:

2-3

  • Marital, Y Vs Job/Education/Housing:

2-7

Continuous variables

  • Account Information
    2-2

  • Economics
    2-14

2-15

Combined

  • Age vs job/education
    2-8

Feature Engineering/Selecting

Lable Encoding:

  • Transform education type into education year:
    2-5

Extracted features:

  • year features and monthly contact features from labelled month:

2-9

2-10

Delete/modify bad feature

  • weekday of last contact:

2-11

  • previous and poutcome are similar, so just keep one.

2-13

Missing value and ouliters handling:

We used Random Forest, Pmm linear prediction,
mode/Average, and combined them.

Modelling

We used Logistics, SVM and decision tree.

Results:

3-1

3-2

GitHub