3.1.1. Naïve Bayes

The name of the NB algorithm is formed by two words, 'Naïve' and 'Bayes'. While the latter word stands for the Bayes (named after Thomas Bayes) theorem, which is used for calculating the conditional probability of the occurrence of landslides, in NB, the first term stands for the assumption that the algorithm naively considers all parameters to be independent of each other. The use of simple Bayes' theorem helps the model to have good mathematical control and the results can be achieved fast by using an NB algorithm [20]. The equation for calculating conditional probability of occurrence of landslide (*L*), subject to the occurrence of conditioning factors *C* (*<sup>C</sup>*1*to Cn*) is given in the following equation:

$$P\left(L \mid \mathbb{C}\_1, \mathbb{C}\_2, \dots, \dots, \mathbb{C}\_n\right) = \frac{P(L) \times P(\mathbb{C}\_1, \mathbb{C}\_2, \dots, \mathbb{C}\_n \mid L)}{P(\mathbb{C}\_1, \mathbb{C}\_2, \dots, \mathbb{C}\_n)}\tag{1}$$

The advantage of an NB algorithm is its simplicity and lower calculation time. The model does not require any hyper parameter tuning and can be easily implemented on any dataset. The major limitation is its assumption of independent parameters. The assumption does not hold true for most of the real-world problems and hence the algorithm may not provide reliable results when the parameters are highly dependent on each other. The algorithm has been used in LSM for more than a decade [21].
