Second-hand mobile phone price prediction algorithm based on deep fusion of time sequence process and mobile phone defect features

Abstract

The invention belongs to the technical field of second-hand mobile phone price prediction, and particularly relates to a second-hand mobile phone price prediction algorithm based on deep fusion of a time sequence process and mobile phone defect features. The method comprises the following steps: extracting metadata features; calculating the average sales price of each machine type every day as a macroscopic time sequence; performing text preprocessing on the content of the model detection report; employing a word2vec model to map each word into one 300-dimensional real vector, and inputting the obtained mobile phone metadata features, the mobile phone price time sequence and the text vector representation into a price prediction model for prediction, wherein the price prediction model comprises a time sequence process modeling module, an attribute feature modeling module and an attention fusion module. According to the method, the global long-term trend of the price and the local sudden fluctuation of the price in unit time are considered in the time sequence process modeling of the second-hand mobile phone price prediction, and the change in historical sales time sequence price data can be more comprehensively and specifically modeled; besides, a CNN with an attention mechanism is used to model sudden fluctuations in price per unit time.

Description

technical field [0001] The invention belongs to the technical field of second-hand mobile phone price prediction, and in particular relates to a second-hand mobile phone price prediction algorithm deeply integrated with time series process and mobile phone defect features. Background technique [0002] The price prediction of second-hand mobile phones is of great significance to the recycling and sales of second-hand mobile phones. It is a regression model to predict the selling price by modeling various characteristics such as the type, age, used time, and geographical location of second-hand mobile phones. A typical application scenario. [0003] However, the prediction of mobile phone prices will be affected by external factors and produce unpredictable short-term fluctuations. The attributes and metadata characteristics of mobile phones usually contain multiple modes, which also increases the complexity of modeling. [0004] Industry technologies can be mainly divided i...

AI Technical Summary

Problems solved by technology

[0011] 1. External factors affecting mobile phone prices may cover different ranges and durations. It is artificially difficult to assume the quantity and shape of price fluctuations. Capturing short-term fluctuations through specific assumptions about external factors in advance or manual extraction will limit the model predictive power

[0012] 2. There is a lack of text modeling of the test reports of quality inspectors. When users buy mobile phones, in addition to paying attention to the test results of the mobile phones, they will also pay attention to the test reports given by the testers, and the writing of the reports (details are omitted) , the level of words used, etc.) will also affect the user's psychological bid

Existing technologies fail to make full use of long text based on inspection reports for second-hand mobile phone price prediction

[0013] 3. Failure to fully integrate temporal process and content feature modeling to leverage their respective strengths

The intuition fusion method lacks the flexibility to deal with the price evolution process

Images