feat: implement RandomForest-based sales forecasting with multi-format data extraction

[VasanSoundararajan]

• Added RandomForestRegressor as the primary model for sales forecasting
  (replacing Linear Regression) to better capture non-linear patterns.
• Implemented extraction functions for CSV, PDF, DOCX, and image files
  (via pdfminer.six, python-docx, and pytesseract).
• Added robust parsing with regex patterns for multiple date formats and sales numbers.
• Introduced CLI interface with commands for handling different file types.
• Added forecast plotting with matplotlib (historical vs predicted sales).
• Improved logging for better traceability of data extraction, parsing, and training steps.

[github-actions]

Thanks for creating a PR for your Issue! ☺️

We'll review it as soon as possible.
In the meantime, please double-check the file changes and ensure that all commits are accurate.

If there are any unresolved review comments, feel free to resolve them. 🙌🏼

[vercel]

@VasanSoundararajan is attempting to deploy a commit to the vaibhav2154's projects Team on Vercel.

A member of the Team first needs to authorize it.

[aleenaharoldpeter]

@VasanSoundararajan Hey! I think it might be cleaner to put the read functions in PyEveryday/tree/main/backend/scripts/data_tools/. That way, you can just call them in your ML code and focus only on the ML logic. @Vaibhav2154 Please correct me if I’m off here!

[aleenaharoldpeter]

Nice work on this refactor, @VasanSoundararajan 🙌, the split definitely makes SalesPredictor cleaner already.

Quick thought (@Vaibhav2154 ): do you think it might be even cleaner if we also moved PDF/DOCX/Image extractors into data_tools, so that all data ingestion lives in one place and SalesPredictor only handles training/prediction logic? That would make the separation of concerns super clear and keep ML logic focused.

That said, Vasan — please don’t make any changes yet 🙏 since @Vaibhav2154 hasn’t had a chance to weigh in. Just wanted to put this out there for discussion once everyone’s had a chance to look.

[VasanSoundararajan]

Ok I will do it sooner @aleenaharoldpeter but i think so the data tool is for all implementing little stuff that can be inside of each files

[aleenaharoldpeter]

Nice work on this refactor, @VasanSoundararajan 🙌, the split definitely makes SalesPredictor cleaner already.

Quick thought (@Vaibhav2154 ): do you think it might be even cleaner if we also moved PDF/DOCX/Image extractors into data_tools, so that all data ingestion lives in one place and SalesPredictor only handles training/prediction logic? That would make the separation of concerns super clear and keep ML logic focused.

That said, Vasan — please don’t make any changes yet 🙏 since @Vaibhav2154 hasn’t had a chance to weigh in. Just wanted to put this out there for discussion once everyone’s had a chance to look.

Ok I will do it sooner

Hey, btw I just noticed I had a normal CSV read in converter and chunked in processor — I’ll refactor that so it’s consistent. So give me a little time on that. And yeah, I think you’re right, just wanted @Vaibhav2154’s input too so we’re all on the same page.