Application research of SGD algorithm based on PCA dimensionality reduction technique for stock price prediction

After decades of development, stock investment plays an increasingly important role in people's life. The stock market is directly related to the stability of the financial market and sustainable development of the economy. In the field of financial investment, with the rise of machine learning in the artificial intelligence, a large number of scholars use machine learning methods to study market changes, industry trends, and predict stock prices, so as to achieve guidance on stock investment. Stochastic gradient Descent (SGD) is an optimization algorithm used to train model parameters. Unlike traditional gradient descent, SGD uses only one sample or a small batch of samples at a time to update the parameters, reducing the computational complexity. It is especially suitable for training large data sets and complex models. This paper selects the daily degree data of Amazon from May 15, 1997 to October 17, 2021, and selects Open, Close, High, Low, Volume, Adj Close as the features. First, PCA is used to reduce the dimension of the data. Then the stochastic gradient descent algorithm was used to fit the training set data, the test set data fitting graph and different evaluation indexes were used to evaluate the prediction accuracy of the model algorithm. Compared with other algorithms, it was found that the prediction model based on this method had high prediction accuracy and small error. The article has certain practical significance which provides reference for relevant researchers.