Advertisement prediction optimization method and device based on machine learning and computer equipment

The invention relates to an advertisement prediction optimization method and device based on machine learning, computer equipment and a storage medium. The method comprises the following steps: firstly, collecting user behavior data, advertisement interaction data and advertisement putting data, and performing cleaning and standardization processing to obtain an accurate initial data set; then, key features are extracted from the initial data set, interaction features between the user and the advertisement are constructed, and an initial feature set is formed; and then, converting the feature set into a data form acceptable to a machine learning algorithm by using an encoder, carrying out model training and parameter adjustment, and evaluating the model performance by using multiple algorithms. Finally, the trained model is applied to new advertisement putting request data, and accurate prediction of the advertisement effect and the user behavior is achieved. According to the scheme, the user behavior data an