UNPKG

cerceis-lib

Version:

Contains list of quality of life functions that is written in TypeScript and es6

github.com/Cerceis/LIB

Cerceis/LIB

33 lines (22 loc) • 1.54 kB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
# K-means-clustering Identify list of values by user specified numbers of group (k). # How does the algorithm works? 1. Specifiy K (Number of groups)(K > 1). 2. Randomly select K numbers of data points within the area (min, max of input data) as ***initial clusters*** 3. Measure distance of data point to **each** cluster point, then assign the data point to the **nearest** cluster point. Then repeat the same step with **each** data point. 4. Calculate the mean of each cluster. 5. Repeat step **3** but instead of cluster point, we measure the distant of each point to the **mean** value we got from step **4** 6. If the cluster does not change, then we are done. Extra: By peforming the algorithm once does not guaratee the perfect clustering, so additionally, ***attempt*** should also be specified. By redoing the K-means multiple times with different ***initial clusters***, it can then determine the best attempt by look at similar sum of count of **unique** values on each result culster. Extra2: If input K is unspecified or unable to determined. The algorithm could auto run K starting from K = 1 untils it find the best variations in each culster. ## Determine best variant To determine the best variant as the result. We will look at the sum of difference of each child and take the lowest value (Most evenly distributed). ``` 6 : 2 : 3 //Numbers of child of each cluster 4,3 4,1 3,1 = 7 + 5 + 4 = 16 4 : 3 : 4 1,0 1,1 0,1 = 1 + 2 + 1 = 4 Lowest the better ``` # Dependencies - CLIB [GetSmallest] - CLIB [RandomInt]