# AC

## Doubts

- TP is a loan not given that was a good decision, since it's the minority class

## Business Understanding

### Business Objectives

The main goal of this project is to provide a good algorithm that detects which future given loans will be paid, and consequentially which ones should be given.

### Assess Situation

- Several databases with an often indirect relationship
- Missing data

### Determine Data Mining Goals

The objective of the implemented algorithm is to serve as an operation blackbox that can be used. Visual explanations will not be the main consideration (since a lot of data will be confusing to see represented as a tree), instead it should be operated as a simple "function call" by a banking clerk/manager.

A bank's purpose is not to lose money, so the bank loans it concedes that don't work out well are more impactful than the ones it doesn't concede (neutral situation, although money can be lost in this situation as well), recall is appropriate.

### Produce Project Plan

- Data Preprocessing/Analysis per table (gather data to diminish the total columns that will be provided to the algorithm)
- Data Processing for business understanding,
- Apply different models and score them accordingly

## Data Understanding

### Collect initial data

### Describe Data

### Explore Data

### Verify Data Quality

[//]: # (BU: analysis of requirements with the end user)

[//]: # (BU: definition of business goals)

[//]: # (BU: translation of business goals into data mining goals)

[//]: # (DU: diversity of statistical methods)

[//]: # (DU: complexity of statistical methods)

[//]: # (DU: interpretation of results of statistical methods)

[//]: # (DU: knowledge extraction from results of statistical methods)

[//]: # (DU: complexity of plots)

[//]: # (DU: presentation)

[//]: # (DU: interpretation of plots)

[//]: # (DU: visual knowledge extraction)

[//]: # (DU: diversity of plots)

[//]: # (Data understanding: basic statistics, univariate, bivariate and multivariate graphs; critical interpretation of the results from statistical and domain
perspective)