Abstract: In the article has been conducted a research aiming increase of classification result stability of commercial bank’s debtor creditworthiness with usage of boosted decision trees algorithm with application of stratified sampling. The general principle of the Kaggle data science research platform is described, in which statistics and data mining specialists compete to create the best models for forecasting and data modelling based on the data offered by companies or users. Has been conducted an analysis of models and program implementation of boosted decision trees algorithm for estimation of commercial bank’s debtor creditworthiness. The most effective program packages are described — XGBoost and LGBM, which are used for program implementation of boosted decision trees. For confirmation of the results, has been used a program package LGBM on data of Home Credit Bank, available in the scope of Home Credit Competition on data science platform Kaggle. The details of Home Credit Competition are shared: conducted a description of input data, a description of an approach for creation of characteristics for training a model and technical approach which was proposed during participation in the competition. During the research proposed to use stratified sampling of control dataset by target variable and the most significant characteristics during training of a model to increase a stability of the result of classification and enhance efficiency during a process of modernization of model’s architecture. Proved experimentally, that the use of stratified sampling of the control sample during the training of boosted decision tree models makes possible to increase the stability of the model result, which increases the efficiency of validation of modernization of the model architecture.