CN-121997145-A - Market signal processing method and system

Abstract

The application relates to the technical field of data processing, in particular to a market signal processing method and system; the method comprises the steps of obtaining a market signal, wherein the market signal at least comprises electronic commerce platform sales data of a target product and social media sound volume data representing the heat of the target product, performing wavelet transformation on the market signal to obtain a wavelet coefficient, calculating a noise threshold, comparing the energy intensity of the wavelet coefficient with the noise threshold to filter noise in the market signal, constructing market state vectors capable of representing market states at all times based on the noise-filtered market signal, and inputting the market state vectors at all times in a preset history period into a pre-trained classifier to obtain the probability that the current market is in all market states. The application has the effect of improving the accuracy of market state identification.

Inventors

• DING ZIJIAN
• DING JUNWEI
• CHEN DEPIN

Assignees

• 钛动科技股份有限公司

Dates

Publication Date

20260508

Application Date

20260207

Claims (10)

1. 1. A market signal processing method, comprising: acquiring market signals, wherein the market signals at least comprise electronic commerce platform sales volume data of a target product and social media sound volume data representing the heat of the target product; calculating a noise threshold value, and comparing the energy intensity of the wavelet coefficient with the noise threshold value to filter noise in the market signal; Constructing market state vectors capable of representing market states at all times based on the noise-filtered market signals, and inputting the market state vectors at all times in a preset history period into a pre-trained classifier to obtain the probability that the current market is in all market states; The market state vector at least comprises a demand growth feature, wherein the demand growth feature is positively correlated with the sales of an electronic commerce platform at the current moment and is positively correlated with the growth rate of forward media sound in social media sound data.
2. 2. A market signal processing method according to claim 1, wherein the forward media volume is a volume containing data with positive rating or purchasing motivation or objective description or regular counseling for the target product.
3. 3. The market signal processing method according to claim 1, wherein calculating the noise threshold includes, for any time, constructing a historical reference sliding window based on the time, determining the noise threshold according to historical statistical characteristics of noise obtained after the market signal in the historical reference sliding window is decomposed and the number of original observation points acquired in the historical reference sliding window at the time, and positively correlating the noise threshold with the number of original observation points in the historical reference sliding window.
4. 4. The method of claim 1, wherein calculating the noise threshold includes extracting a mean value of noise from the reference signal for a preselected specific time period, and taking a sum of standard deviations of the mean value and a preset multiple as the noise threshold corresponding to each time.
5. 5. The method of claim 1, wherein the market signal further comprises a bid price of a bid of the same type as the target product, and wherein the market state vector further comprises a competitive fluctuation feature characterizing fluctuation of the bid price and a cross-channel correlation feature characterizing a varying correlation between e-commerce platform sales and forward media sound.
6. 6. The method for processing market signals according to claim 2, wherein the step of calculating the demand growth characteristics includes, for any moment, obtaining a growth rate of the sales of the e-commerce platform and a growth rate of the forward media sound volume at the moment, and taking a result of weighted summation of the two as the demand growth characteristics.
7. 7. The market signal processing method according to claim 1, wherein the pre-trained classifier is a phase classifier based on a transducer architecture.
8. 8. The market signal processing method according to claim 1, wherein the market state comprises at least two of a demand inflection point state indicating a sudden change in demand of the target product, a plateau indicating a steady demand of the target product, and a decay period indicating continuous atrophy of the market of the target product.
9. 9. The market signal processing method according to claim 1, further comprising, in response to the probability of a market state being greater than a preset threshold and persisting for more than a preset period of time, determining that the current market is in the market state.
10. 10. Market signal processing system, characterized by comprising a processor and a memory, the memory storing computer program instructions which, when executed by the processor, implement a market signal processing method according to any one of claims 1-9.

Description

Market signal processing method and system Technical Field The present application relates to the field of data processing technologies, and in particular, to a market signal processing method and system. Background Along with the development of data-driven marketing and intelligent decision-making technology, the data-driven marketing and intelligent decision-making technology is rapidly developed, and the marketing mode and decision-making flow of enterprises are deeply changed. Enterprises increasingly rely on real-time collection and deep analysis of market signals in carefully formulating marketing strategies. The market signal is like a wind vane in the business world, accurately reflects the dynamic change of the market, and guides the forward direction for enterprises. The market signal has a wide coverage range and comprises but is not limited to multisource time series data such as sales volume change of an e-commerce platform, public opinion enthusiasm in social media, search trend, bid price adjustment condition and the like. The sales volume change of the electronic commerce platform directly reflects the purchase intention and demand intensity of a consumer on a product, is a visual reflection of market demands, the public opinion heat degree in social media reflects the discussion heat degree and emotion tendency of the consumer on brands and products, can help enterprises to know the attitudes and expectations of the consumer, the search tendency can reveal the attention degree of the consumer on specific products or services to indicate potential market demands, the competitive price adjustment condition affects the competition pattern of the market, and the enterprises need to pay close attention to formulate reasonable price strategies. The multi-source time series data are mutually interwoven and mutually influenced to jointly form a complex and changeable market signal system. The method can accurately reflect market demand change, consumer behavior trend and competition pattern evolution to a certain extent, so that the method is applied to marketing decision-making systems, market monitoring systems and automatic marketing platforms and becomes an important basis for enterprises to formulate marketing strategies. In the related art, a processing manner for market signals is mainly focused on trend identification based on statistical analysis or a simple time series model. For example, one common approach is to determine whether the market has a tendency to grow or slip by moving averages, exponential smoothing, or homonymy/cyclic calculations on the historical data. The moving average method eliminates the influence of short-term fluctuation by carrying out average processing on data in a certain period, thereby displaying the long-term trend of the data more clearly, the index smoothing rule gives different weights to the data in different periods, the influence of recent data is more emphasized, the change of the market can be reflected more timely, and the increase or decrease amplitude of the market is intuitively displayed by comparing the same-ratio/ring-ratio calculation with the data in the same period or the previous period. And in another scheme, by setting a fixed threshold, when indexes such as sales volume, click rate or social media sound volume exceed or are lower than a preset threshold, early warning is triggered or decision-making staff is prompted to intervene. The scheme is simple and direct, and can prompt enterprises to pay attention to market change in time when indexes are abnormal. In addition, some systems introduce regression models for periodic analysis or trend fitting of market changes. The regression model predicts future trends of the market by building a relationship model between the variables. However, in practical applications, the market signal itself has significant non-stationarity and high noise characteristics. On one hand, market data is easily influenced by factors such as short-term sales promotion, holiday effect, platform algorithm adjustment, data acquisition delay or abnormality, and the like, so that a large amount of random fluctuation or sporadic abnormality is mixed in signals, and on the other hand, different data sources have differences in sampling frequency, statistical caliber and response time lag, so that market signals in multiple dimensions are difficult to directly align in time dimension and amplitude dimension. Under the above circumstances, it is often difficult to effectively distinguish between an effective signal that truly reflects the structural change of the market and random noise in the existing technical solutions based on simple smoothing or threshold judgment. Further, the prior art generally focuses on the monitoring of "numerical changes" and lacks the ability to identify the stage or evolution state of the market. For example, when a short-term drop in sales of a certain category occurs, existing systems of