Create prediction software based on Excel data

This project involves developing predictive analytics software to forecast horse race outcomes using machine learning techniques. An initial Excel dataset with 20 columns of numeric and text data will serve as the training material, comprising approximately 15,000 rows that will grow almost daily. The aim is to automate analysis of this information to generate accurate predictions, improving on current manual methods achieving 50-60% accuracy. The freelancer will design software to independently examine the data, learn from patterns, and provide forecasts with a single command. As new rows accumulate, the program must self-update its methods to maintain high performance. Within the scope of work, the freelancer will code the application, teach it to comprehend the dataset, and ensure extensibility for continual dataset enlargement. An estimate reflecting the requirements described is requested.