landmines_clustering_training

Model Card for Model ID

This model will use the information about power usage to predict what type of landmine is being described. This uses information from the Land Mines dataset from UC Irvine Machine Learning Repository.

Uses

This model was created for a university assignment and is not ready to be used for production purposes.

How to Get Started with the Model

from huggingface_hub import hf_hub_download
hf_hub_download(repo_id="CSC310-fall25/landmines_clustering_training", filename="model.pkl",local_dir='.')
dt_loaded = sio.load('model.pkl')

Training Data

<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->

Data Source: UCI Machine Learning Repository

Training Procedure

<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->

There were no missing values in this dataset and there was no otlier detection or removal done on the dataset. I selected the features by taking the first two columns with float values, and I removed the column with the target variable from the lm_X set alltogether. There was also no scaling or normalization done to the dataset. I used the KMeans method to cluster this data as it is what we learned in class. The K value 5 was selected because there are 5 mine types described in the data initially. This model was trained using many packages including seaborn, numpy, sklearn, pandas and matplotlib. There are two random seeds used, 1108 and 118.

Model Examination

<!-- Relevant interpretability work for the model goes here -->

Land Mines Data Set:
lm_X size: (338, 3)
V: float64
H: float64
S: float64
lm_y size: (338,1)
M: int64

There is no senstive data nor missing data in this dataset originally.

This data doesn't form very spherical or separated clusters. A lot of the clusters seem to layer on top of eachother in some way. Due to the silhouette score, you may thing that there are well separated and dense clusters, but the ARI and NMI both point to there being no relationship between the predicted and the true labels.

Results

Silhouette Score: 0.324
Adjusted Rand Index (ARI): -0.027
Normalized Mutual Information (NMI): 0.049

Model Card Authors

Brianna Landry

Model Card Contact

Contact Brianna Landry @:
[email protected]
[email protected]
github: briannalandry8