Business process model with “RandomForest” algorithm

Currently, the areas of application machine learning are multifaceted: artificial intelligence, financial applications, bioinformatics, intellectual games, speech and text recognition, computer language processing, medical diagnostics, technical search rubrication. Machine is an area scientific knowledge associated with learning-capable algorithms. The use methods explained by fact that for most intelligent complex tasks (for example, etc.) it almost impossible to develop obvious algorithm their solution. However, you can teach a learn how solve such problems. In our article, we propose model based on “RandomForest”, which allows one recognize bots HTTP sessions. chosen provides many advantages: non-iterative learning; high quality resulting models (comparable neural networks ensembles networks); small number adjustable parameters. It works well missing data (retains good accuracy); internal assessment generalizability model; able work raw without preprocessing. was trained dataset more than 5000 prospects this direction obvious, since robotic traffic accounts 40% total Internet traffic.