Built with logic.Driven by curiosity.

THE PROBLEM

Monthly visitor-arrival and hotel datasets use wide tables and inconsistent values. They need a common date structure before analysis and prediction.

WHAT I BUILT

• Reshape with melt and pivot_table; parse year/month fields and inner-join hotel and arrival data on date.
• Select numeric arrays, impute missing values with medians, remove duplicates and cap outliers using IQR bounds.
• Engineer lag, rolling-average and seasonal features; compare scaling, PCA and eight regression models.

THE PROBLEM

Predict smoker versus non-smoker from physiological measurements while handling ratio-derived invalid values and evaluating across balanced validation folds.

WHAT I BUILT

• Engineer BMI, pulse pressure, lipid ratios, log-transformed enzymes and vision/hearing composites; convert DataFrames into NumPy feature arrays.
• Replace infinities with missing values and impute using training medians. Split with Scikit-Learn’s five-fold StratifiedKFold.
• Train LightGBM, XGBoost and CatBoost with early stopping; collect out-of-fold probabilities, optimise blend weights by ROC-AUC and average test predictions.

THE PROBLEM

A hospital menu needs to collect valid patient information, book appointments, calculate bills and assign triage rooms without failing on incorrect input.

WHAT I BUILT

• Use functions, while loops, conditionals and try/except ValueError to reject empty names, invalid ages and incorrect menu choices.
• Validate department and calendar-date inputs; compute a $100 consultation plus $10 per lab test, with subsidised patients paying 70%.
• Translate pseudocode and flowcharts into severity-based room assignments. Document VS Code/Pylance debugging and input-loop corrections.

THE PROBLEM

Stationery stock changes when items are added, edited or sold. Quantities and total values must remain consistent, and sales must not exceed stock.

WHAT I BUILT

• Define StationeryItem with name, quantity, price and a calculate_total() method; store item objects in a dictionary.
• Add and edit items with numeric validation; reject non-positive sold quantities and sales beyond available stock, then recalculate values.
• Display stock, export inventory through Python’s csv module and calculate highest-value stock, lowest stock, average price and total inventory value.