This repository contains code for K-Means Clustering in Python, based on the work from the following Kaggle notebook: K-Means Clustering with Python by prashant111.
Facebook pages of 10 Thai fashion and cosmetics retail sellers. Posts of a different nature (video, photos, statuses, and links). Engagement metrics consist of comments, shares, and reactions.
[https://archive.ics.uci.edu/dataset/488/facebook+live+sellers+in+thailand]
| Variable Name | Role | Type | Description | Units | Missing Values |
|---|---|---|---|---|---|
| status_id | ID | Integer | No | ||
| status_type | Feature | Categorical | No | ||
| status_published | Feature | Categorical | No | ||
| num_reactions | Feature | Integer | No | ||
| num_comments | Feature | Integer | No | ||
| num_shares | Feature | Binary | No | ||
| num_likes | Feature | Integer | No | ||
| num_loves | Feature | Binary | No | ||
| num_wows | Feature | Binary | No | ||
| num_hahas | Feature | Binary | No | ||
| num_sads | Feature | Binary | No | ||
| num_angrys | Feature | Binary | No |
- In this project I have implemented Kmeans Clustering.
- Use minmax and Standard scale to normalize data and choose minmax for nomralization because clustering result were better through that
- Use Elbow method to find the best number of cluster in Kmeans
- Based on elbow method number of clusteres equals 2 is the best value.
- Analyse the result of Kmeans Clusetring with silhouette method.
- Based on Silhoute method average score equals 0.88
- Analyse clustering by grouping clusters based on cluster's label
- make a visual consideration over Kmeans Clustering with pair plot and scatter plot.