Step 0: Introduction Naive Bayes Theorem
Bayes' theorem is one of the earliest probabilistic inference algorithm, proposed by the Reverend Bayes (his reasoning for the existence of God), in some use cases of the theorem is still useful.
It is best to explain the theorem is through an example. Suppose you are a Secret Service personnel, you received a mandate, it is necessary to protect his / her safety in a particular campaign speech in the Republican presidential candidate. This is a campaign speech that everyone can participate in public activities, your task is not simple, we need to keep in mind the danger exists. One way is for everyone to set a threat factor, according to human characteristics (such as age, sex, carry bag and tension, and so on), you can determine whether there is a threat this person.
If a person meets all of these features, has exceeded the threshold doubts in your mind, you can take steps and the person away from the active site. Principles of Bayes' theorem is true, we will be in accordance with certain related events (someone's age, sex, with or without a package, tension, etc.) to calculate the probability of the occurrence of an event (someone threat) probability.
You also need to consider independence between these features. For example, if in the event there was a child looks very nervous, so nervous compared with adults, children possibility of threat will be lower. To better explain this, look at the following two characteristics: age and stress levels. Suppose we study these features alone, we can design all of a nervous person regarded as a potential threat to people's model. However, it is likely there will be many false positive cases, because the scene of a minor is likely to be nervous. Therefore, taking into account age and "tension" feature will certainly be more accurate reflection of who threats.
This is the meaning of the word of the theorem "simple", this theorem will consider each feature remain independent from each other, but in reality is not always so, and therefore will affect the final conclusion.
In short, Bayesian according to some other event (in this case the information is classified as spam) joint probability distribution is calculated for an event (in this case the information is spam) probability of occurrence. Later we will understand the principles of Bayes' theorem, but at first understand our data to be processed.
