machine learning ✦ VS Code, crisp-dm
This project applies machine learning and logical reasoning to real-world problems. Using the CRISP-DM framework, I prepared data, implemented clustering with kMeans, detected outliers using LOF and ISF, and performed classification with kNN, Decision Trees, and Random Forest. Additionally, I developed a Prolog-based logic system to simulate a robot's coffee-making process through procedural planning. I handled data preprocessing, algorithm implementation, and logical modeling while ensuring clear documentation and effective results presentation.
step 1: Problem Understanding and Objective
The project began with identifying the objectives for two distinct tasks: applying machine learning techniques to analyze real-world data and using logic-based programming for procedural planning. The CRISP-DM task involved clustering data, detecting outliers, and classifying information to derive meaningful insights. Simultaneously, the Prolog task required creating a logical system to simulate a robot's coffee-making process, emphasizing procedural reasoning and goal-oriented planning.
Step 2: Approach and Implementation
For the CRISP-DM task, I cleaned and prepared the dataset, addressing missing values and ensuring usability. I implemented clustering using kMeans to group data, outlier detection with LOF and ISF to identify anomalies, and classification with kNN, Decision Trees, and Random Forest, using feature importance analysis to extract key insights. For the Prolog task, I developed logical axioms to represent states, actions, and preconditions, creating a procedural system that allowed a robot to achieve its coffee-making goal through logical planning.
Step 3: Results and Insights
The CRISP-DM analysis revealed critical features such as hour and motor_vehicle_death_count, which significantly impacted classification models. The Random Forest algorithm stood out for its accuracy and interpretability. In the Prolog task, the robot successfully executed a sequence of logical actions to complete the coffee-making process, showcasing the practical applications of procedural reasoning. These tasks demonstrated my ability to tackle complex challenges using both machine learning and logic-based approaches.
