Welcome,
I use Python primarily as a tool for solving applied problems in data analysis and business contexts — including statistical analysis, machine learning, and modeling. This repository showcases a range of completed and ongoing projects. Each includes relevant datasets and code, covering common use cases such as clustering, time series forecasting, and predictive modeling. I focus on practical, data-driven solutions that may be useful for others working in similar areas.
| Project | Description | Year |
|---|---|---|
| Musical Preferences Analysis | Analysis of user behavior on a music streaming service, with a focus on comparing musical preferences across cities. The analysis identifies which genres are most popular in different locations and how listening patterns vary throughout the week. The project simulates a real-world data analysis task involving data cleaning, filtering, grouping, and aggregation.. Key points: Python, Pandas, functions, loops, data structures, slicing. |
2021 |
| Loan Default Prediction | Investigation of the reliability of borrowers, focusing on how their personal characteristics, such as family status, number of children, and income, influence their ability to repay loans on time. The dataset contains various attributes of bank clients, such as demographic information and loan payment history. The goal is to determine whether factors like marital status, number of children, and income type are significant predictors of loan default. Key points: data pre-processing, descriptive statistics, lemmatization. |
2021 |
| Real Estate Market Analysis | This project analyzes a dataset containing real estate listings for apartments in St. Petersburg and the Leningrad region, focusing on identifying key factors influencing property prices. The primary objective is to explore and preprocess the dataset, clean missing or anomalous values, and ultimately build a model to predict apartment prices based on various features Key points: matplotlib, plot, box plot, scatter plot, query. |
2021 |
| Analysis of mobile tariffs | Data analysis with selection between 2 mobile tariffs: which one generates more revenue depending on customers' behavior and region. Key points: scipy, hypothesis, t-test, p-value. |
2021 |
| Game market analysis | Analysis of game sales within 1980-2016 with best suited genres and platforms for NA, EU and JP markets. Key points: matplotlib, seaborn. |
2021 |
| Best suited mobile tariff | Machine learning model deployment - prediction the best suited mobile tariff to a customer. Key points: data split (train, valid, test), classification models, decision tree, random forest, logistic regression, accuracy score. |
2022 |
| Modeling of customers' attrition | Supervised learning ML models (decision tree, random forest and logistic regression) of cunsumer business attrition (weighting target classes, upsampling and downsampling) reaching sufficient F1. Key points: problem of data disbalance, class weighting, upsampling, downsampling, ROC curve, ROC-AUC, F1 score. |
2022 |
| Researching of oil deposits | Machine learning model for choosing the best region for drilling deployment, based on its oil deposits and business profit Key points: regression models, linear regression, bootstrap, R2. |
2022 |
| Gold flotation ML model | Prediction of gold recovery effectiveness based on technological process. Key points: GridSearchCV, RandomizedSearchCV, KNNImputer, pipeline. |
2022 |
| Protecting customer data | Data encoding with matrix algebra tools. Key points: matrix, linear regression. |
2022 |
| Car cost prediction | Comparison of 3 models (RandomForestRegressor, LightGBM and CatBoost) in terms of the best RMSE and learning/validation time, using ordinal encoding for qualitative attributes. Key points: gradient boosting, lightgbm, CatBoostRegressor, OrdinalEncoder. |
2022 |
| Taxi prediction | Prediction of number of taxi orders within next hour, based on short trends in time series, day of week and hourly seasonality. Models are tested by RMSE. Key points: TimeSeriesSplit, seasonal_decompose, resample, rolling_mean. |
2022 |
| Classification of comments | NLP classification model with preliminary text stemming and vectorization. The model has to classify a user's comment as toxic or non-toxic based on sentiment analysis. Key points: WordNetLemmatizer, TfidfVectorizer, SGDClassifier. |
2022 |
| Project | Description | Year |
|---|---|---|
| Monte Carlo stock price simulation | A Monte Carlo simulation for stock price trajectories using Geometric Brownian Motion. It demonstrates how to model potential future stock prices based on historical volatility and expected returns, providing a visual representation of possible price paths. | 2025 |
| Capacity Planning Optimization | Capacity planning is a critical aspect of operations management that involves determining the optimal capacity required to meet future demand while minimizing costs and maximizing efficiency. It ensures that resources, such as machinery, personnel, and production lines, are utilized effectively to achieve organizational goals. | TBD |
| A/B testing | A Python-based A/B testing framework that analyzes the impact of UI changes on conversion rates. It includes statistical tests (t-test, Mann–Whitney), normality and variance checks, uplift calculation with confidence intervals, and clear visualizations. Designed to support data-driven decision-making in product development. | 2025 |
| Critical Path Method Analizer | A Python tool that calculates the critical path for project schedules using the Critical Path Method (CPM) - one of the most popular tool in project management. It processes task dependencies, computes early/late start/finish times, and visualizes the project network to identify tasks that directly impact the project timeline. | 2025 |
| Project | Description | Year |
|---|---|---|
| Titanic survivals | A classical assignment from Kaggle competition | 2023 |
| Stock price prediction model | A simple machine learning stock price prediction model (D1). | 2024 |
| Project | Description | Year |
|---|---|---|
| Todo bot | A Telegram bot helps to create simple tasks and review user's schedule. | 2021 |
| Pet pic bot | A Telegram bot sends a nice random picture of cat or dog to the user via chat message. | 2023 |
| API bot | A Telegram bot checks status of homework at Yandex.Practicum (Russian e-learning company) via API each 10 minutes and provedes its status to the user as Telegram message. | 2023 |
| Blog | A blog with full functionality (registration, pass reset, searching, posts publication/edition/deletion, comments publication/edition/deletion, images posting etc.) | 2023 |
| Spanish bot | A Telegram bot helps to learn Spanish words. | 2024 |
| Foodgram | A web service for cooking recipes publication. | 2025 |
- LinkedIn: Roman Krasovskiy