Clustering

The structure of a dataset for a clustering algorithm like K-Means is quite simple and can be understood practically. Each row represents a unique sample, and each column contains specific information about that sample. The goal of this type of dataset is for the AI model to learn to identify similarities/differences between the data.

STRUCTURE

1. Rows

   • Represents a sample (an element to be grouped, such as a customer, a product, or any record).

2. Columns (X)

   • They contain the properties or characteristics of the sample, also called features. Each property is a numerical value that will be used by the algorithm to calculate similarities and differences.

EXAMPLE

Imagine you want to group e-commerce customers based on two characteristics:

1. Number of purchases made

2. Total spent.

The dataset would look like this:

Number of purchases made	Total spent
5	300.50
10	1200.00
3	150.75
8	800.00
2	100.00

• Line 1 (header): Column names that represent the properties (optional).

• Subsequent lines: Data for each sample, with values separated by commas.

DATA IMPORT

In all DelphAI objects, it is possible to import the dataset through a CSV file or a TDataset.

1. CSV:

   • The CSV file must follow the same format as the table above.

   • Example of a CSV file:

     ParamA,ParamB
     5,300.50
     10,1200.00
     3,150.75
     8,800.00
     2,100.00

2. TDataset/Query:

   • The dataset can be stored in a relational database.

   • Example of a SELECT query on the table in the database:

     ParamA | ParamB
     -------|--------
     5      | 300.50
     10     | 1200.00
     3      | 150.75
     8      | 800.00
     2      | 100.00

   • Use an SQL query to select the data:

     SELECT * FROM Clients;

RULES AND TIPS FOR CREATING THE DATASET

1. Data Consistency

   • All rows must have the same number of columns.

   • All values in the columns must be in numeric format.

2. No Missing Values:

   • Every cell must have a value (no "gaps" are allowed).

EXAMPLE DATASET

You can find an example of the CSV file in the official repository.