# Credit Score Card - Alibaba Cloud Machine Learning Platform

The article aims at providing a solution to Credit Card scoring on Alibaba Cloud Machine Learning Platform

# Credit Score Card - based on Credit Card Bill Statements on AlibabaCloud Machine Learning Platform

What is ScoreCard?

• A Score Card may be defined as semi - Structured report of an individual that can be used to allotting or accessing some important risk-taking task or decision to improvise some credential.

What is Credit - ScoreCard?

• Credit ScoreCard is a mathematical model that provides a qualitative and quantitative analysis of the probability that the customer will display a defined behavior that might include a lower level of delinquency or loan default or even bankruptcy.
• It is also used in other industries for scoring individual, customer service scoring, and even AliPay credit scoring.

Aim for the article

• This article aims and provides the procedure in Scoring of Credit based upon Credit Card Bill Statements on Alibaba Cloud Machine Learning Platform.

First Step

The first step in any of the execution of the experiment is to gather the datasets, thus we take the datasets from the source: https://www.kaggle.com/uciml/default-of-credit-card-clients-dataset?

Second Step

The Second Step is to create a stable workflow as follows:

Third Step

The Third Step is to Data Split into two parts - one for training and another for testing or prediction of result assessment.

Fourth Step

The fourth step is very important as it focuses on one-hot encoding, defining itself as grouping the input data into data classes. The data values in each class are replaced by the value, which is the representative of the class.

After the process the data looks as follows:

Fifth Step:

Population Stability Index: PSI, in short, is a metric to identify a shift in the population for credit scorecards. By comparing the stability of the population before the data split, after data split, and after data binding, the model created classifies as follows :

Sixth Step:

The sixth step is to train scorecard, the purpose is to normalize scores to indicate the weights of the features in the model as

• Unscaled - it indicates and represents the original weight
• Scaled - it represents the number of points that a feature gains or loses.
• Importance - It indicates the influence of each indicator on the prediction result

Finally, the modeling results as follows for the above experiment:

Reference :