FERNANDO LOIZIDESMAIMONA MOHAMMEDALI ABDULBAQI TURKISTANI2022-06-042021-12-292022-06-04109444https://drepo.sdl.edu.sa/handle/20.500.14154/66095The paper applies a predictive model for the failure of companies based on Companies' House filing. The client of this study is the Tramshed Tech organisation. They wanted to experiment to check if the companies that delay in submitting their annual account are failed companies or not. Corporate failure has been a subject of impressive scholastic premium since the mid-1960s. Some noticed that the corporates' last year before bankruptcy tended to delay reporting their annual accounts. Moreover, in the light of rapid technological development, some experimenters have researched investment and management decision-making using Machine Learning. Thus, this study used the XGBoost algorithm to predict the non-submission account companies' failure. The data is from Companies House. The dataset is 1,564,292 companies. The results are 95% accuracy, 93% f-score and 73% AUC score.38enThesis