Gold Price Prediction

🏅✍️ Gold Price Prediction 🟡💰

Gold has been considered for centuries a store of value, protecting wealth over the long term—especially during periods of inflation, currency depreciation, or economic instability. Central banks hold significant gold reserves, reflecting its importance in the global financial system.

Your role is to develop an automated system capable of predicting the gold closing price (gold close) using a complex set of financial, economic and market indicators, including stock indices, commodity prices and foreign exchange rates.

📁 Dataset

You have the following CSV file available:

• gold.csv – contains historical data from 2010 through 2024 about financial markets, commodities and economic indicators.

Each row contains the following columns:

• date – date in YYYY-MM-DD format
• sp500 open, sp500 high, sp500 low, sp500 close, sp500 volume, sp500 high-low – S&P500 values
• nasdaq open, nasdaq high, nasdaq low, nasdaq close, nasdaq volume, nasdaq high-low – Nasdaq values
• us_rates_% – US reference interest rate
• CPI – Consumer Price Index
• usd_chf, eur_usd – exchange rates
• GDP – Gross Domestic Product
• silver open, silver high, silver low, silver close, silver volume, silver high-low – silver values
• oil open, oil high, oil low, oil close, oil volume, oil high-low – oil values
• platinum open, platinum high, platinum low, platinum close, platinum volume, platinum high-low – platinum values
• palladium open, palladium high, palladium low, palladium close, palladium volume, palladium high-low – palladium values
• gold open, gold high, gold low, gold close, gold volume – gold values (our target is gold close)

💡 The data may have mixed granularities (daily, monthly, quarterly). Some values are missing (NaN), and the model must handle these correctly.

🧠 Task (100 points)

Build a system that can predict the gold closing price (gold close) for the rows in test.csv.

Save your predictions in a file named submission.csv with the following format:

1ID,gold close2R00001,1820.353R00002,1815.924R00003,1822.10

where:

• ID – the unique identifier of the row from test.csv
• gold close – the price predicted by your system

📊 Evaluation

Predictions will be compared against the real values in ground_truth.csv and evaluated using Root Mean Squared Error (RMSE):

1RMSE = sqrt( sum((y_true - y_pred)^2) / n )

Final scoring is based on the RMSE obtained:

• RMSE ≤ 0.5 → 100 points
• RMSE ≥ 5 → 0 points
• For intermediate values, the score is assigned proportionally between 0 and 100.