# Predicting Seismic Hazards with Machine Learning & MLP
> Learn how Machine Learning and Deep Learning models like KNN and MLP are used for seismic hazard prediction in mining using the Seismic-Bumps dataset.

Tags: machine-learning, deep-learning, seismic-hazard, neural-networks, data-science, mining-safety, mlp, knn
## Seismic Hazard Prediction via Machine Learning
* Analysis of mining seismic bumps using sensor data.
* Comparison of Unsupervised (K-Means), Supervised (KNN), and Deep Learning (MLP) models.

## Dataset Overview: "Seismic-Bumps"
* **Source:** UCI Machine Learning Repository.
* **Volume:** 2584 records with 10 features like seismic energy (genergy) and pulse counts (gpuls).
* **Challenge:** High class imbalance between hazardous and non-hazardous states.

## Methodology & Preprocessing
* **Label Encoding:** Converting categorical shifts (Morning/Afternoon) to numerical values.
* **RobustScaler:** Using Median and IQR to handle extreme energy outliers instead of standard scaling.

## Model Implementations
* **K-Means Clustering:** Grouping data into 3 clusters based on energy/pulses to find natural risk regimes.
* **KNN Algorithm:** Risk determination based on 5 neighbors using Euclidean distance; used as a baseline.
* **MLP Architecture:** Neural network with 10 input features, hidden layers (32 and 16 neurons) with ReLU activation, and a Sigmoid output for probability.

## Stabilization & Training
* **Dropout (0.2):** Randomly disabling 20% of neurons to prevent overfitting.
* **EarlyStopping:** Halting training when validation loss stops improving.

## Results & Performance
* **Correlation:** Spearman Correlation of ~0.76 between seismic energy and pulse counts.
* **Accuracy:** KNN reached ~91% accuracy, while MLP (Deep Learning) achieved ~93.5%.
* **Conclusion:** Deep Learning models show the most promise for early warning systems due to their ability to capture non-linear relationships in seismic data.
---
This presentation was created with [Bobr AI](https://bobr.ai) — an AI presentation generator.