Project Name: FindYourStocks - Stock Recommendation System

Developers: Leisha Murthy and Rohan Bhalla

Problem to Solve:

Provide investing beginners with a convenient way of looking up companies (stocks) to invest in based on their interests and preferences, taking into account industries and stock performance.

Statistical Methods and Machine Learning Algorithms Used:

• Content-Based Filtering Approach: Utilized statistical methods like CountVectorizer and machine learning algorithms such as cosine similarity.
• CountVectorizer: Converted textual stock data into a numerical representation for easy processing.
• Cosine Similarity: Measured the similarity between stocks based on their content, helping recommend the most similar stocks to users.
• Model Evaluation: Employed quantitative methods like cross-validation and performance metrics, alongside qualitative checks against industry data sites to validate predictions.

Other Potential Models Considered:

• TF-IDF: Was evaluated but not chosen due to dataset limitations.
• Linear Approach with Direct Comparison: Rejected for not addressing dataset nuances effectively.

Business Applications:

• Personal Investment Firms and Financial Advisors: Can use the model to offer personalized stock recommendations to clients, improving investment decisions and returns.
• Individual Investors: Can leverage the model to identify and evaluate potential investments based on their preferences and interests.
• Insight into Investor Sentiment: The model can offer insights into investor sentiment, aiding various business decisions.

Possible System Extensions:

• Recommendations Beyond S&P 500: The system's framework can be adapted to recommend stocks outside of the S&P 500.
• Extended Dataframes for Different Timeframes: Addition of dataframes for 3-year, 5-year, and 10-year stock trend data with averages and percent change information, aiding in recommending stocks based on users' short-term or long-term investment strategies.