Development of IPPO: Integrated Predictive Printing Order and Inventory Management System

Vocabulary practice flashcards covering the architecture, machine learning models, database structure, workflows, and evaluation methodology of the IPPO system for SLN Printing.

IPPO

An acronym for Integrated Predictive Printing Order and Inventory Management System, a centralized web-based platform developed for SLN Printing that automates orders, inventory, point of sale transactions, production workflows, and machine learning-based demand forecasting.

SLN Printing

A small-scale printing and merchandise business located at 925 Lerma Street, Sampaloc, Manila, operated by one owner-manager and five staff members, serving as the client and study setting for IPPO.

Three-Tier Architecture

A software architecture pattern that separates the system into three distinct layers: the Presentation Layer (responsive frontend), Application Layer (business logic and machine learning engine), and Data Layer (centralized relational database).

Presentation Layer

The client-facing tier of IPPO consisting of responsive HTML5, CSS3, and plain JavaScript interfaces accessible via desktop and mobile web browsers, displaying inline SVG charts without external charting libraries.

Application Layer

The server-side tier of IPPO built with Python $3.14$, Django $5.2$, and Django REST Framework, handling URL routing, role-based authentication, business logic, email dispatch via Gmail SMTP, and machine learning forecasting via scikit-learn, pandas, and NumPy.

Data Layer

The centralized data tier of IPPO powered by a MySQL $8.0$ relational database using the utf8mb4 character set, storing all customer records, orders, inventory, production workflows, sales transactions, and demand forecasts.

Bill of Materials (BOM)

The defined specification mapping each product to the exact quantities of raw materials it consumes (e.g., blank shirts, ink, transfer film), used to automatically deduct inventory upon order approval and convert historical sales into material usage.

Order Management Module

The functional module enabling customers to place orders online with custom specifications or design files, while allowing administrators and staff to approve, reject, monitor, and update order statuses in real time.

Inventory Management Module

The module that tracks real-time stock levels of printing materials, automatically subtracts stock upon order approval and POS completion, raises configurable low-stock alerts, and provides decision support via reorder points and economic order quantities.

Point of Sale (POS) and Payments Module

The module that processes walk-in and pre-order transactions via Cash, GCash, Maya, and Xendit online checkout, generates itemized receipts, and immediately synchronizes sales with centralized inventory records.

Workflow Tracking Module

The component that monitors active order production across four sequential stages: Queued, In Production, Quality Check, and Ready for Pickup or Delivery, preventing stage skipping and alerting staff to bottlenecks.

Machine Learning-Based Demand Forecasting Module

The analytical engine that converts historical orders into weekly material usage and evaluates Random Forest Regression against Linear Regression via time-series cross-validation to predict four-week material requirements and detect seasonal peaks.

Reorder Point (ROP)

The stock inventory threshold calculated as $\text{ROP} = \text{average daily use} \times \text{supplier lead time (days)} + \text{safety stock}$, triggering a 'Restock' recommendation when on-hand quantity falls at or below this level.

Economic Order Quantity (EOQ)

The inventory management formula $\text{EOQ} = \sqrt{\frac{2 \times D \times S}{H}}$ used to find the optimal replenishment quantity, where $D$ is annual forecast demand, $S$ is order placement cost ($\text{₱}500$), and $H$ is annual holding cost ($20\%$ of unit cost).

Safety Stock

The inventory cushion computed as $1.28 \times \text{forecast uncertainty}$, utilizing a $z$-value of $1.28$ to provide approximately a $90\%$ service level against unexpected demand variations during supplier lead times.

Random Forest Regressor (IPPO Configuration)

An ensemble regression model configured with $200$ decision trees ($\text{n\_estimators}=200$), a maximum depth of $8$, a minimum of $2$ weeks per leaf, and random seed $42$, kept intentionally shallow to prevent overfitting on small datasets.

Time-Series Cross-Validation

A five-fold model validation method that evaluates models chronologically across historical splits without shuffling, ensuring models only train on past data and predict unseen future weeks to prevent data leakage.

Root Mean Square Error (RMSE)

The model selection metric computed as $\text{RMSE} = \sqrt{\frac{1}{n} \sum (\text{actual} - \text{predicted})^2}$, used as the primary decider between Random Forest and Linear Regression because it heavily penalizes large demand misses.

Mean Absolute Error (MAE)

A secondary evaluation metric computed as $\text{MAE} = \frac{1}{n} \sum |\text{actual} - \text{predicted}|$, representing the average weekly magnitude of forecast errors in raw material units.

Weighted Absolute Percentage Error (WAPE)

A forecast error metric defined as the total absolute difference between actual and predicted usage divided by total actual usage, reported in IPPO backtesting as an accuracy score of $100\% - \text{WAPE}$.

Feature Engineering (Clues)

The predictive input variables generated from historical material usage, consisting of lag features (lag 1, 2, 3), rolling averages and standard deviations over 3 and 5 weeks, calendar markers (week of year, month), and a binary peak-season flag for June, July, October, and November.

Role-Based Access Control (RBAC)

A server-side security mechanism enforcing distinct system privileges across three user roles: Administrator (full access, analytics, settings), Staff (POS, production tracking, order updates, stock view), and Customer (browsing, ordering, tracking own records).

PBKDF2 (Password-Based Key Derivation Function 2)

The cryptographic one-way password hashing algorithm employed by Django within IPPO to hash user passwords, preventing plain text storage in the database.

Xendit

The third-party Philippine payment gateway integrated into IPPO, providing a secure online checkout page that processes GCash, Maya, and credit/debit card transactions with server-side payment verification.

Acknowledgment Receipt

The formal label applied to sales transaction receipts generated by IPPO, reflecting that official Bureau of Internal Revenue (BIR) certification and accreditation are outside the study's scope.

Waterfall SDLC Model

The sequential software engineering lifecycle chosen for IPPO comprising six distinct phases: Planning, Requirements Gathering, System Design, Implementation, Integration and Testing, and Deployment and Maintenance.

Total Population Sampling

A non-probability sampling technique where all members of a defined population ($16$ total individuals: $1$ owner/administrator, $5$ staff members, and $10$ customers) are included in system evaluation due to the small user base.

ISO/IEC 25010

The international standard for software product quality used to evaluate IPPO across seven specific characteristics: Functional Suitability, Performance Efficiency, Compatibility, Usability, Security, Maintainability, and Portability (with Reliability verified via system testing).

Key Informant Interview (KII)

A qualitative requirement gathering and evaluation method involving structured, in-depth interviews with the business owner and operational staff at SLN Printing to identify bottlenecks and validate system features.

What system model does this diagram represent?

The Level 0 Context Diagram of IPPO, showing the system interacting with four external entities: Customer, Staff, Administrator (Owner), and the Xendit payment gateway.

What software architecture diagram does this figure depict?

The Three-Tier Architecture and deployment diagram of IPPO, illustrating client requests entering via HTTPS through Nginx and Gunicorn to Django and MySQL.

What development process model is shown in this diagram?

The sequential Waterfall Software Development Life Cycle (SDLC) model utilized for the development and testing of IPPO.

What architectural lifecycle diagram is represented here?

The Order and Production Lifecycle of IPPO, detailing state transitions from Pending, Approved, In Production, Quality Check, Ready for Pickup, to Completed.

What data design model does this diagram illustrate?

The Entity Relationship Diagram (ERD) of IPPO, outlining the relational connections and foreign keys among core tables such as Customer, User, Order, Inventory, and Forecast.