Challenges I Ran Into:

Scoping and Data Acquisition:

• Defining the Scope: Initially, I spent time figuring out the best approach to tackle the problem statement, considering various options and limitations.
• Dataset Challenges: Finding readily available datasets was difficult. I had to source the data myself, which proved to be a significant hurdle.
• Time Constrains (Phase2): Enchancing the project based on discussed things with mentor & trying to properly scope & best possible way to implement it.

Data Collection and Processing:
To ensure privacy, our data scraping process excludes personally identifiable information like names and profile photos. We only extract review text, ratings, owner responses & similar non-sensitive data.

• Scraping Data: I scraped reviews from Google for major electronics retailers like Croma, Reliance Digital, Viveks, and Vasanth & Co.
• Data Cleaning and Formatting: I used Python Pandas to clean, extract latitude-longitude (lat-lng) coordinates, and convert the data into JSON format.

Integrating with GenAI:

• Prompt Engineering: I created multiple prompts to be used with the Gemini API, iteratively processing the review data for each branch, split by rating and month, while staying within the free tier's rate limits.
• Bundling and Splitting Data: I bundled the data by branch, then further split it by star ratings (1-5) and months for efficient processing with the Gemini API.

Decision Making and Visualization:

• Prioritization: The biggest challenge was deciding what to focus on, what data sources to use, and what to leave out.
• Visualization Complexity: Visualizing the location markers and implementing supporting algorithms took a considerable amount of time and effort.

Overcoming the Hurdles:

• Breaking Down Tasks: I broke down the project into smaller, manageable chunks, which helped me progress steadily and avoid feeling overwhelmed.