HDB Flat Price Forecasting System using GPT-4o and Time Series Analytics - n8n Workflow

Real estate investors and analysts needing automated property price forecasts.
Data scientists looking for advanced n8n templates integrating AI agents and custom statistical tools.
Users who require monthly, scheduled data collection and predictive modeling.
Anyone needing to automatically save AI-driven predictions into Google Sheets.

This sophisticated n8n workflow acts as an AI-powered real estate analyst. It solves the challenge of manually aggregating, cleaning, and analyzing vast amounts of historical data required for accurate price forecasting. The n8n workflow automatically fetches multi-year HDB resale data, performs deep statistical and time-series analysis (calculating moving averages, volatility, and seasonal patterns), and feeds these features into an OpenAI GPT-4o agent. By leveraging a custom statistical forecasting tool within the agent, the system generates probabilistic 12-month predictions with clear confidence intervals. This automated process ensures timely, high-quality, and data-driven insights, making this a powerful n8n template for finance and property professionals.

The entire n8n workflow operates monthly via the Monthly Data Collection Trigger n8n trigger.

1. Data Collection: The Workflow Configuration node sets up necessary parameters. Four concurrent HTTP Request n8n node components fetch the current year's data and three years of historical records from the HDB public API.


2. Consolidation & Cleaning: The Merge All Historical Data n8n node combines the datasets. The subsequent Data Cleaning and Normalization n8n node uses custom JavaScript to standardize field formats, remove outliers, and handle duplicates, ensuring clean input for analysis.


3. Feature Extraction: Two parallel Code n8n node components—Statistical Pattern Mining and Time Series Analysis—extract key predictive features, including correlations, YoY growth rates, momentum, and seasonality.


4. AI Synthesis: The Aggregate Statistical Features n8n node combines these features. The AI Forecasting Agent uses the powerful OpenAI GPT-4 Model and the specialized Statistical Forecasting Tool to synthesize these inputs and generate the future price forecast, complete with confidence bounds derived from Monte Carlo simulation.


5. Output: The Structured Forecast Output Parser ensures the prediction is machine-readable. Finally, the Format Forecast Results n8n node adds essential metadata, and the Save to Google Sheets n8n node automatically logs the final, structured 12-month predictions generated by the n8n workflow.

To use this n8n template, follow these steps:

1. Import: Import the provided n8n workflow JSON into your n8n instance.


2. OpenAI Credential: Set up your OpenAI API credentials in the OpenAI GPT-4 Model n8n node. This requires access to the GPT-4o model.


3. Data Resource ID: In the Workflow Configuration n8n node, replace the placeholder PLACEHOLDERVALUEHDB Resale Flat Prices Resource ID with the actual resource ID from the data source API.


4. Google Sheets Setup:

1. Activate: Set the Monthly Data Collection Trigger n8n trigger to active and save the n8n workflow.

Monthly Data Collection Trigger (n8n trigger): Initiates the n8n workflow monthly, scheduled for 2 AM, ensuring regular, automated data refreshes.
HTTP Request (n8n node): Used multiple times to fetch raw HDB resale price data from the Singapore government API, retrieving current and historical years in parallel for efficiency.
Data Cleaning and Normalization (Code n8n node): Crucial for data integrity. Standardizes town names, validates price and area fields, and removes duplicate entries before analysis.
Statistical Pattern Mining (Code n8n node): Performs grouped statistical analysis (mean, median, correlation) and calculates Year-over-Year price growth rates based on the historical data in the n8n workflow.
Time Series Analysis (Code n8n node): Generates technical indicators like 3/6/12-month Moving Averages, volatility metrics, and seasonal indices to identify underlying trends.
AI Forecasting Agent (Langchain Agent n8n node): The core intelligence. It uses the GPT-4o model to process aggregated features and decides when to execute the statistical tool for complex calculations. This is a key component of this advanced n8n workflow.
Statistical Forecasting Tool (Code Tool n8n node): A custom tool for the agent that executes advanced models (Linear Regression, Monte Carlo Simulation) to produce numerical 12-month forecasts with confidence intervals.
Save to Google Sheets (Google Sheets n8n node): Final step for persistence. Appends the structured forecast (including predicted price and confidence bounds) into a designated Google Sheet, completing the n8n workflow.