Time Series Prediction via Image Conversion and CNN Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting future trends in time series data, such as stock market trends, often result in low accuracy due to the lack of clear patterns, with many approaches struggling to perform better than random guessing, especially when patterns or trends are not easily recognizable.
Innovation Solution
The method involves converting time series data into images, using convolutional neural networks for feature extraction, and then applying image similarity methods to predict future trends by identifying similar past data patterns, thereby reducing the complexity of data preparation and improving prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional numerical data processing methods are used for time series prediction, then the data preparation process is straightforward, but prediction accuracy is low when clear patterns are absent
Solution Approach 1:
The patent introduces an intermediary representation layer by converting time series data into image format. This intermediary representation allows the application of powerful image recognition algorithms (CNNs) to time series data, thereby improving prediction accuracy while the automated nature of the conversion reduces manual data preparation complexity
Solution Approach 2:
The patent replaces traditional mechanical numerical data processing methods with image-based processing. By transforming time series data into visual representations, the system leverages image recognition capabilities to detect patterns that are not easily recognizable in numerical form, thus improving prediction accuracy
2Reliability
If complex pre-processing steps are applied to time series data, then data quality may improve, but the complexity and time required for data preparation increases
Solution Approach 1:
The patent performs preliminary action by automatically converting time series data into image format as the first step of the process. This pre-processing transformation is done systematically and automatically, ensuring data quality improvement while minimizing the time investment required, as the conversion is a straightforward computational operation
Solution Approach 2:
The patent creates a copy of the time series data in a different format (image representation). This copying process preserves the essential information while transforming it into a more analyzable form, improving data quality without requiring time-consuming manual processing or complex transformations
Data Source
AI summary
A computer-implemented method of generating, from time-series data, a time-series of data sets for predictive analysis, comprises dividing the time-series data into evenly-sized overlapping segments of data, generating an image representing data for each segment, using the time-series data to determine a trend associated with each image, and storing each of the generated images and its associated trend as a data set. In some embodiments of the method the image from each stored data set is transformed into numerical vectors through a feature extraction process using a pre-trained convolutional neural network. The numerical vectors are stored in association with the data set, and the data sets and associated numerical vectors are used to predict the trend for a new time-series image which has been generated from any time-series data.