Sales Forecasting with Python Machine Learning
・Skills: Python, Machine Learning, Pandas
・Overview: A project that automates sales forecasting using Python machine learning.
Python Templates for Digital Marketing Report
Portfolio
In this project, I built a modular, Python‑driven reporting suite that automatically turns Google Analytics 4 (GA4) and Google Search Console (GSC) exports into polished, stakeholder‑ready reports—eliminating manual spreadsheet work and repetitive slide‑building.
View Github RepositoryBackground and Objectives
- Issue: Pulling GA4 and GSC metrics each month often means copy‑pasting CSVs, wrangling formulas, and formatting slides—work that steals hours from actual analysis.
- Solution: My goal was to compress that workflow to a single click: mount Drive, run the notebook, and receive three PDF reports (channel, page group, keyword) that mirror the questions executives ask most.
Main Features
| Template | Questions Answered | Outputs |
|---|---|---|
| GA4 Channel Report | Which traffic sources are growing or shrinking? | sample |
| GSC Page Group Report | How are language or content clusters performing? | sample |
| GSC Keyword Report | Which queries drive clicks—and how do we rank? | sample |
- Instant deltas: automatic vs. Last Month & vs. Last Year columns for every metric.
- AI translation: top Japanese queries are translated to English via LangChain + GPT‑4o, so global teams can follow along.
- PDF compilation: cover page, auto‑generated TOC, side‑by‑side tables & charts, all embedded with Japanese fonts for bilingual clarity.
Use Cases
- Monthly KPI Review: Replace slide decks with auto‑generated PDFs sent to leadership.
- Content Gap Analysis: Spot under‑performing page groups or keywords in seconds.
- International SEO Alignment: Share AI‑translated keyword trends with non‑Japanese stakeholders.
Technical Stack
- Data: Google Sheets + gspread (raw GA4 & GSC exports)
- Data Processing: Pandas, NumPy, fiscal‑year logic in Python
- Visualization: Matplotlib + japanize‑matplotlib for Japanese labels
- AI Translation: LangChain + GPT-4o-mini (OpenAI)
- Report Generation: ReportLab with GenShin Gothic fonts → multi‑section PDF
- Execution: Runs locally or in Google Colab; Drive mounted for seamless file I/O
