Self-directed project
Data Analytics and AI
Data Science · July 2026
A ProoV case study · educational project, not employment
Built an end-to-end demand forecasting workflow on 9,648 historical transactions: performed sales, regional, and product EDA; decomposed monthly revenue to inspect trend and seasonality; and compared a naive moving-average baseline, linear regression, and SARIMA using a leakage-free train/test split. The final comparison showed SARIMA as the best model across MAE, RMSE, and MAPE, supporting more reliable peak-season planning.
Graded against
- Exploratory Data Analysis20%
- Time-Series Analysis25%
- Forecasting Model Building35%
- Business Communication20%
Passed · pass mark 60/100
What stood out6
- Used a chronological 18/6 month split instead of a random split, avoiding leakage
- Compared three forecasting approaches and reported MAE, RMSE, and MAPE for each
- Applied seasonal decomposition and moving averages to inspect trend and annual seasonality
- Built a clear model comparison table showing SARIMA as the strongest performer
- Leakage-Free Time-Series Splitting
- Comparative Forecast Evaluation
The work I submitted20 tasks
12 tasks · 9.4k characters · Python · 8 written answers · 436 words
- Eda Code 1 LoadPython · 1.2k characters
- Eda Step 1 Load242 characters
- Eda Code 1 Describe1.5k characters
- Eda Step 1 Describe246 characters
- Eda Code 1 Timeseries879 characters
- Eda Step 1 Timeseries250 characters
- Eda Code 1 Regional1.1k characters
- Eda Step 1 Regional259 characters
- Eda Code 1 Product1.5k characters
- Ts Code 2 Resample1.2k characters
- Ts Step 2 Resample232 characters
- Ts Code 2 Decompose880 characters
- Eda Code 1 Load90 words
- Eda Step 1 Load28 words
- Eda Code 1 Describe93 words
- Eda Step 1 Describe26 words
- Eda Code 1 Timeseries65 words
- Eda Step 1 Timeseries28 words
- Eda Code 1 Regional79 words
- Eda Step 1 Regional27 words
Summarised on purpose — the submitted code and writing stay private so this page cannot be reused as an answer key. The full submission sits behind the verified certificate.