Landslide Machine Learning

The research objective of this proposal is to figuring out the reasoning of large number of fatalities on specific regions.

This project contains 3 parts.

Understanding the factors that lead to huge number of fatalities from landslide_size
Classifying regions by clustering. (Could be done from using countries in the dataset, but I want to take a step farther)
Visualizing the result by making an HTML that contains a world map and dots for each landslides. (The regions are distinguishable based on colors and size of the dots are based on the number of fatalities for that landslide.)

Part 1 will be supervised regression.

Part 2 will be unsupervised clustering.

Part 3 will be using the html already made from Landslide_visualization.

Digging the number of fatalities at multi-angle (population, landslide_size, distribution, clusters) for better analysis.
Modifying the latitude and longitude calculation while clustering (considering that earth is a sphere)
Find the best number of clusters using kmeans.score() (so far I just set the number of cluster as 5) [DONE]
Increasing the size of the dataset
Adding scatter plot of the clustering for visualization [DONE]

Name		Name	Last commit message	Last commit date
Latest commit History 22 Commits
Part_0_dataset_exploration.ipynb		Part_0_dataset_exploration.ipynb
Part_0_dataset_modification.ipynb		Part_0_dataset_modification.ipynb
README.md		README.md
global_landslides.csv		global_landslides.csv
modified_dataset.csv		modified_dataset.csv
part_1_correlation_of_fatalaties.ipynb		part_1_correlation_of_fatalaties.ipynb
part_2_KMeans_clustering(n=5).ipynb		part_2_KMeans_clustering(n=5).ipynb
part_2_finding_right_number_of_clusters.ipynb		part_2_finding_right_number_of_clusters.ipynb