This repository for Machine Learning Course Homeworks in Bioinformatics Institute 2023
For Homework7 we created custom convolution function and tried to reproduce AlexNet architecture with five convolutional layers and a 2D max pooling for feature extraction. For classification stage, we used two linear fully connected (FC) layers. ReLU activation function was used for both purposes. Different convolution, activation and optimization was performed using pytorch Python package. For data visualization we used matplotlib and seaborn packages.
For Homework6 we tried to build MLP (multiplayer perceptron) with three multiconnected layers and different activation functions and optimizators using torch Python package. For data visualization we used matplotlib packages.
Here some solutions based on CatBoostClassifier from catboost Python package for BI_2023 kaggle Dota2 Competition.
For Homework4 we tried to implement machine learning algorithms, such as Decision Tree, Random Forest, CatBOOST and many others, using custom fuctions for RandomForest Classification, and from scikit-learn, xgboost, lightgbm, catboost Python packages. This task were coded with basic Python and pandas package. For data visualization we used matplotlib and seaborn packages.
For Homework3 we tried to implement Clustering machine learning algorithms, such as KMeans, Agglomerative Clustering, and DBSCAN, using custom fuctions for Kmeans, and from scikit-learn Python package. This task were coded with basic Python and pandas package. For data visualization we used matplotlib, seaborn, and IPython packages.
For Homework2 we tried to implement Linear regression machine learning algorithms using custom fuctions for regression and binary classification, and using scikit-learn Python package. This task were coded with basic Python and pandas package. For data visualization, VIF and QQ-plot creation we used matplotlib, seaborn, and statsmodels packages.
For Homework1 we tried to implement k-Nearest Neighbors machine learning algorithm using custom fuctions for classification and model metrics, and comparing their performance with respective functions from scikit-learn Python package. This task were coded with basic Python and pandas package. For data visualization we used matplotlib and seaborn packages.