The Analysis of The Daily Return Percentage as An Alternative To The Closing Price Of The Stock Using The Ensemble Model

Abstract

Abstract—In stock exchanges, buyers and sellers meet to trade shares in public companies. Stock exchanges encourage investment. Raising capital enables companies to grow, expand, and create jobs in the economy. These investments are critical enablers of trade, economic growth, and prosperity. Data preprocessing and cleaning are integral to machine learning research and studies. This paper highlights the common mistakes present among the various models for stock price prediction, and it also offers the solution to the problem and a procedure to prepare the data, which any model can use as input. Our objective is to comprehend various models and studies that have been normalizing the price data of the stock, which causes the price to be fixed in each range, which is not valid in real-world scenarios; hence, such models will produce inaccurate predictions over a period. To solve this problem, this paper discusses Daily Return % as an alternative to close prices. To the best of our knowledge, this is the first paper discussing the issue of scaling the parameter known as the price. To achieve this, we use exploratory data analysis techniques and data visualization that focus on modeling and knowledge discovery for predictive rather than just descriptive processing. It assures good data quality and consistency. This study examines the issue of data cleaning and pinpoints a possible inaccuracy in a TATA POWER dataset. A quick summary of the current data-cleaning approaches and a survey and assessment of the many perspectives on data cleaning are provided.