Business response method and device based on unsupervised learning

The invention provides a business response method and device based on unsupervised learning. The business response method based on unsupervised learning comprises the following steps: acquiring historical order information of a plurality of entities in a preset historical time period; determining multi-dimensional features of each entity according to the historical order information of the plurality of entities; clustering the multi-dimensional features of each entity to obtain a plurality of entity clusters and corresponding cluster mean vectors; according to a preset feature weight, carrying out weighted summation on the feature value of each feature in the cluster mean vector to obtain a vector parameter corresponding to each entity cluster; layering each entity cluster according to the size of the vector parameter corresponding to each entity cluster to obtain a layering result; and responding to the service request based on the layering result. According to the method, the multi-dimensional features of t