Title: Diamond Price Prediction Using

ARIMA Model

Abstract
The prediction of diamond prices is crucial for investors, traders, and the
jewelry
industry. This study implements the ARIMA (AutoRegressive Integrated
Moving Average) model to forecast future diamond prices based on
historical data. The model analyzes past trends and provides insights into
future market behavior. The experimental results validate the model's
effectiveness in time series forecasting, demonstrating its capability to
predict diamond price fluctuations accurately.

1. Introduction
Diamonds are among the most valuable commodities in the global market,
and their prices fluctuate due to various economic factors. Accurately
predicting diamond prices can help traders and investors make informed
decisions. Time series forecasting
methods, particularly ARIMA, have been widely used in financial and
commodity price prediction due to their ability to model temporal
dependencies. This paper aims to implement and evaluate an ARIMA model
for forecasting diamond prices over the next decade using historical price
data.

2. Time Series Model

Time series analysis involves studying past values to identify patterns and
make future predictions. The ARIMA model is a popular statistical method
for time series forecasting, characterized by three main components:

Autoregressive (AR) Component: Captures the relationship between current

and past values.

Integrated (I) Component: Accounts for differencing to ensure stationarity.

Moving Average (MA) Component: Models the dependency of an

observation on residual errors.

The ARIMA model is denoted as ARIMA(p, d, q),

where: p is the number of lag observations in

the AR model.

d is the number of times differencing is applied to make the series

stationary. q is the size of the moving average window.
3. Proposed Work
The proposed work involves implementing the ARIMA model for predicting
diamond prices based on historical data. The steps include

Data Collection: Historical diamond price data is obtained from reliable

sources.

Data Preprocessing: The dataset is cleaned, formatted, and transformed into

a time series format.

Model Selection: The ARIMA model parameters (p, d, q) are optimized.

Model Training and Evaluation: The model is trained on past data and
evaluated for accuracy.

Prediction and Visualization: The trained model forecasts future prices, which
are visualized using graphs and statistical summaries.

4. ARIMA Model Implementation

The ARIMA model was implemented using Python with the statsmodels
library. The key steps include:

Loading the dataset and ensuring a proper

datetime index. Selecting the target variable for

forecasting.

Fitting the ARIMA model with optimal parameters (p=1, d=1,

q=1). Generating future predictions for the next ten years.

Visualizing the results using line graphs, residual plots, and a pie chart to
analyze the forecasted values.

5. The Meaning of p, d, and q in ARIMA

Model
The parameter p represents the order of the autoregressive (AR) term,
meaning the number of past values used as predictors. The d parameter
indicates the number of differencing steps required to make the time series
stationary, ensuring that predictors are independent. If the time series is
already stationary, no differencing (d = 0) is
needed. The q parameter defines the order of the moving average (MA)
component, which models dependencies based on past forecast errors.
These three parameters together help in defining the ARIMA model for
effective forecasting.
• ARIMA Models are specified by three order parameters: (p, d, q), where
as,

• p is the order of the AR term

 • q is the order of the MA term

• d is the number of differencing required to make the time series

stationary

6. AR and MA Models
An AutoRegressive (AR) model predicts future values using past observations,
making it a regression model based on its own lags. The equation consists
of past values of Y,
coefficients estimated by the model, and an intercept term. A Moving
Average (MA) model relies on lagged forecast errors, meaning past
prediction errors influence future predictions. The equation includes error
terms from previous lags, helping the model correct deviations over time.
An ARIMA model combines both AR and MA components while ensuring
stationarity through differencing, making it a comprehensive approach for
forecasting.

AR model¶

• An Auto Regressive (AR) model is one where Yt depends only on its

own lags.

• That is, Yt is a function of the lags of Yt. It is depicted by the

following equation - l

where as,

• Yt−1Yt−1 is the lag1 of the series,

• β1 is the coefficient of lag1 that the model estimates, and

• α is the intercept term, also estimated by the

model. MA model¶

• Likewise a Moving Average (MA) model is one where Yt depends

only on the lagged forecast errors. It is depicted by the following
equation -

• where the error terms are the errors of the autoregressive

models of the respective lags.

ARIMA model in words:¶

Predicted Yt = Constant + Linear combination Lags of Y (upto p lags) +

Linear Combination of Lagged forecast errors (upto q lags)

Residual Analysis: Residual errors were exa5. Experimental Results

The ARIMA model was trained using historical diamond prices. The model
performance was assessed based on:
Graphical Analysis: The actual vs. predicted values were plotted to visualize
model accuracy.mined to ensure minimal deviation from the true values.

Future Forecast: The model projected diamond prices for the next ten years,
demonstrating a consistent trend aligned with historical patterns.

6.1 Output and Graphical Representations

Actual vs. Forecasted Prices:

A line graph was generated to compare actual diamond prices with

predicted values over time.

Residual Errors Analysis:

The residuals were plotted to assess the model’s accuracy. The residuals
were centered around zero, indicating minimal error.

Pie Chart of Forecasted Prices:

A pie chart was used to represent the distribution of predicted diamond

prices over the next ten years.

Forecasted Prices Table:

The ARIMA model provided the following forecast for the next ten

years: Year

Forecasted Price (USD)

2025

4500.23

2026

4702.45
2027

4920.67

2028

5153.89

2029

5402.12

2030

5665.34

2031

5943.56

2032

6236.78

2033

6545.00
2034

6868.22

7. Conclusion
This study successfully implemented an ARIMA model to forecast diamond
prices. The results confirm that ARIMA is an effective time series
forecasting method, capturing price trends with reasonable accuracy.
Future work may involve integrating machine learning techniques or
external economic indicators to enhance prediction accuracy further. This
research provides valuable insights for stakeholders in the diamond
industry, enabling informed decision-making based on predictive analytics.

8. References

Box, G. E., C Jenkins, G. M. (1976). Time series analysis: Forecasting and

control.

Hyndman, R. J., C Athanasopoulos, G. (2018). Forecasting: Principles and

practice.

Chatfield, C. (2003). The analysis of time series: An introduction.

Hamilton, J. D. (1994). Time series analysis.