Training_Testing_Points
In the context of landslide susceptibility mapping, training sampled points and testing sampled points are used in the process of developing and evaluating landslide susceptibility models. Here's what they mean
Training Sampled Points: Training sampled points refer to the locations or points on the ground where landslides have occurred or have been observed. These points are selected based on historical records, field surveys, or other sources of landslide data. The purpose of training sampled points is to provide data for model training. These points establish the relationship between landslide occurrences and various influencing factors or variables, such as slope, soil type, geology, land cover, rainfall, etc. The training sampled points are typically assigned a binary label indicating whether a landslide event has occurred at that location (e.g., landslide = 1, non-landslide = 0). These labeled points are then used to train the landslide susceptibility model.
Testing Sampled Points: Testing sampled points refer to separate points or locations distinct from the training sampled points. These points are also associated with the same variables used in the training phase. The purpose of testing sampled points is to evaluate the performance and accuracy of the trained landslide susceptibility model. These points assess how well the model can predict the likelihood of landslides in areas where landslide events have not been observed or documented. The testing sampled points validate the model by comparing the predicted susceptibility values with the actual occurrence or absence of landslides at these locations. This evaluation helps assess the reliability and effectiveness of the model in predicting landslide susceptibility. By utilizing both training and testing sampled points, landslide susceptibility mapping aims to develop models that can effectively identify areas with a higher probability of experiencing landslides. The training sampled points are used to establish the relationships between landslide occurrences and the influencing factors. In contrast, the testing sampled points provide an independent dataset to assess the model's predictive performance. This process helps generate reliable landslide susceptibility maps that can assist in risk assessment and management.