Week of September 16, 2024

• Began detailed discussions on potential senior project ideas, focusing on financial analysis, natural language processing, and automation tools.
• Conducted an initial literature review on capital forecasting models and identified opportunities to provide a differentiated product by targeting NASDAQ companies.
• Held team brainstorming sessions to finalize project direction, deciding to focus on capital forecasting for stock market companies.
• Outlined desired features, including financial performance prediction, user login functionality, and detailed insights into quarterly reports.
• Met with potential users and financial experts to validate our project idea, incorporating their feedback into refining the project scope.

Week of September 23, 2024

• Finalized the name "Plutos" for our senior project.
• Defined the project scope and objectives: develop a forecasting model, preview quarterly reports, and ensure user-friendly access to insights.
• Conducted competitor analysis to identify gaps in existing solutions, noting opportunities to set ourselves apart by providing detailed quarterly previews.
• Established preliminary goals for the coming weeks, including setting up the project infrastructure and beginning data collection efforts.
• Discussed possible machine learning models for use in capital forecasting, deciding to explore models like Prophet, LSTM, and ensemble learning methods.

Week of September 30, 2024

• Officially kicked off the senior project and agreed to focus on financial analysis and support, specifically capital forecasting for stock market companies.
• Set the initial scope and finalized the main objectives of the project.
• Assigned team members to specific roles to ensure efficiency and defined three main subgroups: feature selection and scope definition, data collection and preprocessing, and model training and evaluation.
• Started gathering datasets for analysis, including historical financial data of NASDAQ-listed companies.
• Contacted Professor Altay Güvenir to approve Eren Biri as our innovation expert, who agreed to provide valuable insights into forecasting and model evaluation.

Week of October 7, 2024

• Set up GitHub Pages for the Plutos Equities website and started organizing its structure.
• Created sections like founders, about, and project reports to give a professional look to our initial web presence.
• Purchased the domain "plutosequities.com" to establish an online presence and ensure branding consistency.
• Discussed website content to clearly communicate our project's value proposition to users and stakeholders.
• Began drafting potential data sources that we could use for training our forecasting models.

Week of October 14, 2024

• Decided to use Tailwind CSS for styling the website to ensure flexibility, scalability, and a clean visual appeal.
• Added an "under construction" page to plutosequities.com to inform visitors of upcoming features and keep them engaged.
• Created mockups of the planned website layout and discussed different design elements that could enhance user experience.
• Shifted project scope from data scraping to focusing solely on capital forecasting for the top 100 companies on NASDAQ.
• Set up initial website pages, including the founders page, project reports page, and about page.

Week of October 21, 2024

• Provided a detailed R script for Lab 5, focusing on linear model assumptions. This helped establish a foundational understanding of the models we would use in our forecasting.
• Updated the website footer to remove the Bilkent Engineering Building address and adjusted the email position to match our branding preferences.
• Discussed different forecasting models, ultimately deciding to proceed with Prophet for initial capital forecasting due to its simplicity and effectiveness for time series data.
• Continued working on the website, defining separate pages for founders, contact information, and project reports.
• Addressed formatting issues with the footer for consistency across all pages to enhance user experience.
• Started collecting financial data from Yahoo Finance and other sources to support our capital forecasting.

Week of October 28, 2024

• Analyzed competitors in the capital forecasting and financial analysis domain to identify areas where Plutos could differentiate.
• Noted that many existing solutions do not provide previews of quarterly reports with the level of detail we aim to deliver, which gives us a unique selling point.
• Held a team meeting to discuss findings from competitor analysis and brainstorm additional features to set our product apart.
• Updated HTML and CSS code for the website to improve presentation and usability, ensuring a better user interface.
• Considered switching to a different hosting provider instead of GitHub Pages, as we needed more flexibility and customization options.
• Updated the favicon to include our company logo, enhancing our branding.
• Added a "stay tuned" message on the homepage with a background image to create anticipation for future visitors.
• Began drafting the initial architecture of our forecasting model, focusing on how different data sources will feed into the model and support accurate predictions.

Week of November 4, 2024

• Finalized the initial website structure and added content to each section, focusing on founders, project overview, and contact information.
• Began preprocessing the collected financial data, removing outliers and dealing with missing values to ensure the data was clean for model training.
• Created a timeline for model training and evaluation, with milestones for each major step.
• Started working on the data collection scripts to automate the process of updating financial data periodically.
• Held a discussion about the evaluation metrics we should use for the forecasting model, deciding on metrics like RMSE and MAPE for accuracy.

Week of November 11, 2024

• Implemented the Prophet model for initial capital forecasting on a sample dataset to evaluate its feasibility.
• Analyzed the results from the first iteration of the Prophet model and noted areas of improvement, including adjusting parameters for better accuracy.
• Continued developing data collection scripts to ensure consistency and reliability in the data retrieval process.
• Worked on integrating the forecasting results into the website, focusing on visualizations that are easy to understand for end users.
• Updated the "under construction" page with more information about the upcoming features to keep visitors informed.

Week of November 18, 2024

• Improved the preprocessing pipeline based on the insights gained from the initial Prophet model run, ensuring better data quality for subsequent model training.
• Finalized the integration of forecasting visualizations into the website, adding interactive graphs that show predicted financial metrics for the top NASDAQ companies.
• Created documentation for the data preprocessing and modeling pipeline to ensure that all team members are on the same page and can contribute effectively.
• Began researching additional models to potentially incorporate into the forecasting ensemble, including LSTM and ARIMA, for improved performance.
• Held a team meeting to discuss feedback received from initial project stakeholders and identified further enhancements to improve usability and accuracy.

Week of November 25, 2024

• Refined the Prophet model by integrating additional financial indicators and adjusting hyperparameters for better predictive accuracy.
• Enhanced the website with interactive tutorials on how to use forecasting tools, aimed at improving user engagement.
• Reviewed the preprocessing pipeline to ensure it aligns with new data sources, particularly real-time financial news streams.
• Organized a team meeting to finalize the metrics for success for Q1 2025 predictions, emphasizing MAPE and coverage criteria.

Week of December 2, 2024

• Continued work on integrating LSTM models, focusing on preprocessing and testing sample datasets.
• Addressed data reliability issues in preprocessing by implementing additional checks for outliers and inconsistencies.
• Conducted initial testing of the website’s data visualization tools with sample user groups, collecting feedback for improvements.
• Started developing a user guide for Plutos Equities, focusing on forecasting methodologies and interpretation of predictions.
• Updated the “Stay Tuned” section of the website with progress reports to maintain interest from early visitors.

Week of December 9, 2024

• Expanded the dataset to include additional unstructured data sources, such as earnings call transcripts, to enrich model input.
• Improved backend efficiency by optimizing the data ingestion pipeline, significantly reducing data retrieval time.
• Presented project progress to stakeholders, highlighting milestones achieved and remaining tasks for the quarter.
• Continued work on user interface enhancements, incorporating feedback from the prior week’s testing.
• Focused on fine‑tuning Prophet’s parameters for better seasonal adjustments based on historical financial data.

Week of December 16, 2024

• Continued to explore ways to combine Prophet and LSTM models to address diverse financial forecasting needs.
• Held a brainstorming session to discuss long‑term monetization strategies for the platform, focusing on subscription models.
• Worked on finalizing the web application’s scalability tests to ensure performance under high user loads.
• Started documenting the model architecture and data processing pipeline for inclusion in the final project report.
• Prepared for the upcoming presentation by rehearsing key points and reviewing project specifications to ensure alignment with objectives.

Week of December 23, 2024

• Formed project team and clarified individual responsibilities and roles.
• Finalized detailed project roadmap and milestones.
• Conducted initial exploratory data analysis on financial data from NASDAQ companies.
• Set up initial environment for forecasting models (Prophet, LSTM).

Week of December 30, 2024

• Created comprehensive documentation structure.
• Defined explicit task responsibilities for each team member.
• Outlined detailed project roadmap, including clear short‑term and long‑term goals.
• Researched potential forecasting models and evaluated their applicability.

Week of January 6, 2025

• Extensive literature review and research into forecasting methodologies (Prophet, ARIMA, LSTM).
• Explored comprehensive data sources such as EDGAR, Yahoo Finance, and Kaggle.
• Developed preliminary data acquisition scripts.

Week of January 13, 2025

• Implemented parsing scripts for SEC reports.
• Developed and tested financial time‑series preprocessing pipeline.
• Identified key financial indicators for quarterly forecasting.

Week of January 20, 2025

• Executed initial tests using Prophet model on selected tickers.
• Evaluated initial performance and documented key findings.
• Analyzed historical financial indicators relevant for forecasting.

Week of January 27, 2025

• Clearly defined forecasting model targets (EPS, revenue, profit).
• Developed and revised comprehensive project pitch presentation slides.
• Held internal review sessions to ensure accuracy of documentation.

Week of February 3, 2025

• Officially implemented baseline Prophet forecasting model.
• Conducted performance benchmarking using RMSE and MAE metrics.
• Performed detailed competitor analysis (AlphaSense, Sentieo).
• Documented findings and integrated feedback from initial evaluations.

Week of February 10, 2025

• Completed data cleaning and segmentation of financial reports.
• Developed classifiers to determine the relevance of text chunks to financial predictions.
• Conducted preliminary sentiment analysis using BERT models.

Week of February 17, 2025

• Enhanced sentiment analysis workflows with Phi‑4 embeddings.
• Tested and validated relevance classification approaches.
• Coordinated backend development goals with modeling outcomes.

Week of February 24, 2025

• Initiated advanced relevance classifier using BERT models.
• Integrated and visualized EPS forecasts alongside sentiment analysis results.
• Refined data handling procedures.

Week of March 3, 2025

• Integrated macroeconomic indicators (e.g., inflation, interest rates).
• Experimented with LightGBM and Random Forest models.
• Evaluated and documented model improvements and limitations.

Week of March 10, 2025

• Developed a refined scoring system for sentiment‑based relevance.
• Prototyped combined forecasting ensemble using LSTM and Prophet.
• Held team reviews for ensuring alignment of project components.

Week of March 17, 2025

• Developed and deployed a custom annotation tool for financial text chunks.
• Optimized preprocessing pipeline for EDGAR data.
• Conducted frontend milestone review and adjustments.

Week of March 24, 2025

• Fully deployed sentiment‑augmented forecasting pipeline.
• Revised model evaluation metrics (e.g., SMAPE, MAE).
• Collected user feedback for iterative improvements.

Week of March 31, 2025

• Established inference endpoints for streamlined forecasting predictions.
• Supported frontend integration of forecasting insights.
• Updated and refined website content, particularly the “What We Offer” section.

Week of April 7, 2025

• Conducted detailed analysis on misclassified text chunks.
• Tuned and optimized sentiment classification thresholds.
• Improved GitHub Pages deployment and site layout.

Week of April 14, 2025

• Enhanced multi‑quarter forecasting model accuracy.
• Added detailed prediction confidence metrics.
• Organized and managed GitHub repository structure effectively.

Week of April 21, 2025

• Finalized user dashboard visualizations (EPS, sentiment polarity).
• Documented model interpretability and explainability methods.
• Enhanced HTML/CSS for polished visual presentation.

Week of April 28, 2025

• Completed comprehensive documentation and final report.
• Conducted final project presentation to stakeholders.
• Delivered all codebases, presentations, and supporting documentation.