This Repository containing portfolio of Machine Learning combined domains projects completed by me for academic, and self learning purposes. ML-based projects are in the form of Jupyter notebooks, and Python project are .py files. The Python Projects using different types of libraries, models, data structures, functions and techniques.
- Install dependencies using requirements.txt (if available in project folder) otherwise required libraries are mentioned at the start of each notebook.
- Run notebooks as usual by using a jupyter notebook server, colab etc.
- Dataset or downloadable link is mentioned in each project folder.
- House Price Prediction : This project employs Linear Regression and Random Forest regression models for the prediction of house prices and dataset is available at kaggle. It further enhances model performance through feature engineering and utilizes GridSearchCV for optimal hyperparameter tuning, ensuring accurate and robust price predictions.
- Handwritten Digit Recognition using NN and SVM : This project aimed to develop a versatile handwritten digit recognition system using two different machine learning approaches: a Neural Network (NN) and a Support Vector Machine (SVM). The objective was to recognize and classify images of handwritten digits with high accuracy and NN predict the higher accuracy.
- Movie Recommendation System : This project delve into the world of recommendation systems, which utilize algorithms to offer personalized suggestions to users, with a focus on content-based and collaborative filtering approaches.
- Human Activity Recognition with Smartphones : This project is about classifying the human activities based on the smartphone data. The dataset is downloaded from kaggle. Data consist of various features and activitives. In this project used classification algorithims and also apply feature selection methods.
- ML-Based Stock Market Prediction : This project aims to predict tomorrow's S&P 500 index price by utilizing historical data. It begins by downloading S&P 500 price data through the 'yfinance' package, proceeds with data cleaning and preprocessing using pandas, and ultimately prepares the dataset for machine learning applications.
- Customer Segmentation using k-mean CLustering : This project demonstrates the application of unsupervised learning, specifically K-Means clustering, to segment customers based on their annual income and spending ratio. The dataset used for this analysis is obtained from kaggle and consists of customer information.
- Ghosts in the Text: Author Attribution of Spooky Literature : This project involves applying NLP techniques to classify text snippets into three different authors: EAP, H.P. HPL, and MWS. The dataset, sourced from Kaggle, consists of text passages from various spooky and mysterious literary works. The goal is to develop and compare machine learning models for authorship attribution, utilizing techniques such as data preprocessing, tokenization, vectorization (using TF-IDF and Word2Vec), and various classification algorithms (including Logistic Regression, Support Vector Machine, Random Forest, and Naive Bayes). The project aims to identify the most effective approach for accurately attributing authorship to spooky texts, providing insights into the underlying patterns in the writing styles of different authors in the horror genre. The evaluation metrics, including accuracy, precision, recall, and F1-score, will be employed to assess the performance of each model, and the results will be visualized for easy interpretation and comparison.
- Text classification using GaussianNB, and Random Forest : This project implemented different machine learning models for text classsification on the IMDB Dataset of 50K Movie Reviews. Text pre-processing implementation includes (missing values, duplicates, tags removal, lower case etc) and then implemented the GaussianNB, and random forest with number of differnet techniques to increase the accuracy.
- Sentiment_Analysis_using Traditional Apporach and RoBERTa : In this project a Sentiment Analysis classifier with NLTK's traditional and Transformer model approach is implemented. The project is to classify the sentiments of amazon customer reviews. The dataset being used in this project is available at kaggle. The result comparision of tradictional approach and transformer-based model roberta is presented in the project.
- Rock Paper and Scissor Game : The Game includes a simple logic encapsulated in an unconditional while loop, encompassing key steps of user input retrieval, validation, and random computer input generation. The game's progression is defined by a concise if-elif-else statement that enforces the game rules.
- The Minesweeper Game : This Python-based mine detection game, also known as Minesweeper, challenges players to uncover hidden mines within a grid of cells. Clicking on a mine results in a loss, while numbers displayed in cells reveal the count of adjacent mines. To aid in gameplay, users can mark potential mines with a right-click, creating an engaging puzzle experience. Tools: tkinter library
- Text-Based Game : This project is an interactive quiz game where users start by running the program and entering their details. They are introduced to the game's rules in the main menu. Players make decisions with 'Y' for yes or 'N' for no, answering questions by entering choices. If the choice selected by the player is right then player will score 1 point and if its wrong then -1.