Machine Learning Tracks

Two short boot camps organized by Hive Helsinki introducing students to AI and machine learning. Key ML concepts were covered such as linear and logistic regression, data preprocessing, feature engineering, and model evaluation.

Modules

Simple Linear Regression 01

• Visualize the relationship between years of experience and salary using pandas and seaborn;
• Train a linear regression model using scikit-learn on the salary dataset to predict salaries based on years of experience.

Simple Linear Regression 02

• Visualize the relationship between years of experience and salary using numpy and matplotlib;
• Train a linear regression model using gradient descent algorithm, then visualize the results using matplotlib;
• Calculate cost function values over iterations, visualize the cost reduction, and predict salaries based on years of experience.

Multiple Linear Regression

• Load, explore and understand a more complexed car price dataset;
• Clean data by extracting and correcting car company names, ensuring the dataset is more reliable and ready for analysis and modeling;
• Visualize car prices across various categorical variables and outliers using seaborn, gaining insight into how different categorical attributes impact the pricing of cars;
• Perform data augmentation by dividing the car prices into buckets based on predefined ranges and create new features using one-hot encoding to enhance the dataset for modeling purposes;
• Split data into training and testing sets, scale the features using MinMaxScaler, train a linear regression model, and evaluate its performance using R² score.

Binary Classification with Logistic Regression

• Build a logistic regression model and make predictions;
• Evaluate the model's performance using the confusion matrix, and visualize the results using a heatmap.

Multinomial Logistic Regression

• Perform exploratory data analysis with a correlation heatmap;
• Preprocess the dataset by splitting, scaling, and training a polytomous logistic regression model;
• Make predictions, assess its performance using techniques like the confusion matrix, and present the results visually using a heatmap.

Classification With Other Models

• Preprocess a dataset for fraud detection, using data visualization techniques to analyze fraud scenarios and class imbalances;
• Handle class imbalance using SMOTE and scales features;
• Train multiple machine learning models, including logistic regression, KNN, decision trees, random forest, gradient boosting, and XGBoost;
• Evaluate and compare the accuracy of each model to predict fraudulent transactions.