There are provided an apparatus and a method for compressing data, an apparatus and a method for analyzing data and a data management system, which are capable of compressing huge data and accurately reproducing the characteristics of the original data from the compressed data.The data compressing apparatus includes detection means for detecting a multiplicity of data sets, each including n parameter values that vary according to an operation of an object, where n is a natural number; and data compressing means for compressing the data sets by inputting the data sets into an n-dimensional space, arranging neurons smaller in number than the data sets in the n-dimensional space, carrying out unsupervised learning for a neural network on the neurons, and converting the data sets into a neuron model parameter characterizing a neuron model obtained by the unsupervised learning.