We already know the basic concepts of the probability and random variables of an event
If we want to find out the regularity of a random variable, we must know its probability distribution, the probability distribution knows, then everything can be calculated. or retreat to the second, to the small to know its digital characteristics, such as mathematical expectations, variance, and so on. And our ideal is to know the overall, but the reality can not, we retreat and seek the second, to find samples, but know that the sample is not our purpose, our purpose is to know the overall, we must keep this in mind, so we have to study how to get the overall sample you want.
One.
We grab a probability sample of the statistic, which is a sample of one, it can determine the amount of knowledge.
The first type of digital characteristics
1. Such as sample mean, sample variance, mean variance (sample standard deviation), second-order center moment, coefficient of variation, sample moment
2. From a cloth point of view, such as Skewness, kurtosis
The second category (it's my own)
The above numeric characteristics are an estimate of the corresponding overall eigenvalue, but it is more suitable for data analysis that is being distributed. If the overall distribution is unknown, or is not a normal distribution, but a serious bias, or there are some extreme values, the above analysis method is not suitable, but should calculate the median, quantile, three-mean, very poor digital characteristics, the calculation of the above statistical values will be used in order statistics.
Two. With the digital characteristics, it is best to make the distribution, so we have a comprehensive description of the overall situation of the data, we need to study the distribution of experimental data.
The method of describing the test data is mainly
Frequency distribution table, histogram, experience distribution function, QQ graph, stem leaf diagram, Box line diagram
Please look at 2. MATLAB implementation:
1. Descriptive statistical analysis