STA457 Time Series Analysis
Winter 2025
Final Project Guidline
1 Introduction
Modelling commodity prices is essential in various industries, including agriculture, finance,
and trade. However, such modelling presents significant challenges, as commodity prices are
influenced by multiple factors and often exhibit characteristics such as seasonality, volatility,
and external shocks. These complexities make accurate forecasting a difficult but valuable
task.
Cocoa, a key commodity in the global market, plays a crucial role in chocolate production
and other food products. Its price is shaped by various factors, including weather conditions,
geopolitical events, supply chain disruptions, and macroeconomic trends. Understanding and
forecasting its price movements is of great interest to producers, traders, and policymakers.
In this project, you will apply time series analysis techniques to model and predict cocoa
futures prices. You will explore various statistical methods to analyze historical price trends,
assess model performance, and interpret the results. Through this project, you will gain
hands-on experience working with real-world time series data and develop skills relevant to
financial and economic forecasting.
You will have considerable flexibility in this project. There are no restrictions on
the methods you choose—you are encouraged to explore and apply any techniques you find
appropriate, even those beyond the scope of our course materials (This is an excellent oppor-
tunity for you to engage in self-study and further independent research while deepen your
understanding). However, you must provide a clear justification for your chosen approach in
your report. Additionally, aside from the data I provide, you may incorporate any supple-
mentary data you deem relevant. This is a real-world, open-ended problem with no single
correct answer. Your goal is to develop a well-reasoned and data-driven model for forecasting
cocoa futures prices. The assessment criteria will be detailed in the rubric section.
Additionally, this is a problem I have personally encountered. About five years ago, while
working in Singapore, I provided consulting services to a Singapore-based chocolate pro-
duction company. One of their key concerns was understanding and predicting cocoa price
fluctuations to optimize their procurement strategies. This experience highlighted the im-
1
portance and difficulty of modelling commodity prices effectively. I want you to step into
my role at that time, tackling the same challenge I faced. This will give you a hands-on
opportunity to apply time series analysis in a real-world business scenario.
2 Background Information
Cocoa is an essential agricultural commodity and the primary ingredient in chocolate pro-
duction. It is derived from the seeds of the cacao tree (Theobroma cacao), which thrives
in tropical climates with high humidity and consistent rainfall. The production of cocoa
is highly sensitive to environmental and economic conditions, making price fluctuations a
common challenge in the global market.
2.1 Cocoa Production and Growing Conditions
Cocoa trees require warm temperatures (between 20°C and 30°C), regular rainfall, and
well-drained, nutrient-rich soil. They typically take three to five years to reach maturity
and begin producing cocoa pods. The trees bear fruit continuously throughout the year,
but major harvests usually occur in two main seasons—one peak and one mid-crop sea-
son—depending on the region.
2.2 Major Cocoa-Producing Regions
Cocoa is primarily grown in the equatorial belt between 10 degrees north and south of
the equator, where climatic conditions are ideal. The top cocoa-producing countries
include:
• West Africa – The region accounts for over 70% of global cocoa production, with
Ivory Coast and Ghana being the largest producers, followed by Nigeria and
Cameroon.
• Latin America – Countries such as Ecuador, Brazil, and Peru also contribute
significantly to global cocoa production.
• Southeast Asia – Indonesia is the leading producer in this region, followed by
Malaysia and Papua New Guinea.
2.3 Factors Affecting Cocoa Prices
Cocoa prices are influenced by a variety of factors, including:
1. Weather Conditions – Adverse weather, such as droughts or excessive rainfall, can
impact cocoa yields. For example, El Nin˜o events can lead to dry conditions that
reduce production, while heavy rains can increase the spread of plant diseases.
2. Pests and Diseases – Cocoa trees are vulnerable to diseases such as black pod disease
and pests like cocoa pod borers, which can significantly reduce output.
2
3. Supply Chain and Political Stability – Disruptions in key cocoa-producing coun-
tries, such as labor strikes, political instability, or export restrictions, can create supply
shortages and drive up prices.
4. Global Demand – Rising demand from the chocolate industry, particularly in emerg-
ing markets like China and India, affects cocoa prices. Seasonal demand fluctuations,
such as increased chocolate consumption during holidays, also contribute to price vari-
ability.
5. Currency Exchange Rates – Since cocoa is traded in U.S. dollars, fluctuations in
currency values, particularly in producing countries, can impact global cocoa prices.
6. Speculation and Market Dynamics – Cocoa futures contracts are actively traded
on commodity exchanges, and speculative trading activities by investors can influence
short-term price movements.
Understanding these factors is crucial for modelling cocoa prices effectively, as they introduce
various patterns, trends, and external shocks into time series data.
3 Data
In this project, you will work with two primary datasets: cocoa futures price data from the
International Cocoa Organization (ICCO) and climate data (temperature and precipitation)
from Ghana, sourced from the National Centers for Environmental Information (NCEI).
• Cocoa Futures Price Data:
The file “Daily Prices ICCO.csv” contains cocoa futures price obtained from the Inter-
national Cocoa Organization (ICCO). This dataset contains daily closing prices
for cocoa futures contracts traded on major commodity exchanges. The time span of
the data is from 1994.3.10 to 2025.2.27.
• Climate Data from Ghana:
The file “Ghana data.csv” contains climate data in Ghana, the largest cocoa-producing
country in the world. The data is obtained from the National Centers for Environmen-
tal Information (NCEI) include:
– STATION: The id of the observation station
– NAME: The name of the observation station
– DATE: The date of the observation
– PRCP: The daily perception. If blank, then it means there is no perception on
that day.
– TAVG: The daily average temperature observed at 2 meters above ground
– TMAX: The maximum daily temperature
– TMIN: The minimum daily temperature
3
While these two datasets form the core of the analysis, you are encouraged to incorporate
any additional data sources that you find relevant. For example, macroeconomic indica-
tors, currency exchange rates, or other climate-related variables could further enhance your
model. However, any additional data should be properly documented and justified in your
report.
4 Output Requirements
Your final output for this project should be a well-structured report that clearly presents your
approach to modelling cocoa futures prices. The report should be written in a formal style
and include appropriate explanations, justifications, and interpretations of results. Below is
a suggested structure for your report:
1. Introduction The introduction should provide an overview of the project, including a
brief description of the study, the motivation of the study and its real-world significance.
You may also briefly outline the objectives of your analysis and the key challenges
involved.
2. Literature Review This section should summarize relevant research on time series
forecasting and commodity price modelling. Highlight how your approach builds upon
or differs from existing methods.
3. Methodology Clearly describe the forecasting methods you choose to apply. As
mentioned earlier, this may include classical time series models covered in this course,
or more sophisticated, e.g., machine learning approaches. Justify your choice of models
based on theoretical considerations, past research and/or the pattern of the data.
Explain in detail any steps, including preprocessing steps such as handling missing
data, stationarity checks, transformation or feature engineering.
4. Data Provide a detailed description of the datasets used in the analysis. For any
external data, please indicate the source of the data. You may also include Summary
statistics and visualizations that help illustrate the characteristics of the data.
5. Forecasting and Results Present the results of your forecasting models, including:
• Model training and validation process.
• Performance evaluation using appropriate metrics.
• Forecasted values and any observed patterns.
• Graphical representations of predictions versus actual prices.
6. Discussion and Conclusion Interpret your results and discuss their implications.
You may also consider including the limitations of your approach, the challenges you
face and how can your analysis be improved or extended.
7. Appendix
4
The appendix should include any supplementary materials that support your analysis
but are not suitable for the main body of the report. The full code used for the project
must be provided in the Appendix. The appendix may also include: Additional elab-
orations, derivations, or mathematical justifications and extended statistical analyses,
tables, or figures that complement the main discussion.
8. References Include a properly formatted list of references for any external sources
cited in your report. Follow a standard citation style (e.g., APA, IEEE, or any format
specified in the submission guidelines). All literature, datasets, and tools used in your
analysis should be appropriately credited.
Your report should be well-organized and written in a clear and concise manner. The use
of formulas, tables, figures, and references is encouraged to support your analysis and they
should be well labelled. Ensure that your submission adheres to proper citation standards.
5 Marking Rubric
As mentioned in the course syllabus, the assessment of the group project is divided into two
components: 20 marks for the overall quality of the project and 10 marks for individual
contributions. For the first 20 marks, all group members will receive the same score, and a
detailed rubric will be provided as follows. The remaining 10 marks will be awarded based
on peer evaluations, reflecting each individual’s contribution to the project.
The first 20 marks for the overall quality of the project will be assessed based on three
key criteria: Writing and Organization, Method Correctness, and Creativity. Below is a
breakdown of how each section will be evaluated.
1. Writing and Organization (5 marks, 25%) – Clear and concise writing that effec-
tively communicates ideas. – Well-organized structure with appropriate use of headings,
subheadings, and sections. – Logical flow of content with smooth transitions between
sections. – Proper spelling, grammar, and formatting throughout the report. – Appro-
priate use of tables, figures, and visualizations to support the analysis.
2. Method Correctness (5 marks, 25%) – Appropriateness of the chosen methods
and models for the problem with accurate application. – Justification of model choices,
with clear explanations of why certain methods were used. – Proper handling of data,
including necessary preprocessing steps. – Clear presentation of the models.
3. Prediction Performance (5 marks, 25%) – Accuracy of the forecasts based on
relevant performance metrics (e.g., RMSE, MAE, MAPE). – Interpretation of forecast
results and discussion of their reliability.
4. Creativity (5 marks, 25%) – Originality and creativity in the approach to the
problem. – Exploration of techniques beyond the course material, including any self-
driven research. – Innovative solutions or novel insights derived from the data and
analysis. – Effective use of external data or additional sources to enhance the model
or analysis. – Thoughtfulness in addressing the limitations of the chosen methods and
5
suggesting improvements or future work.
6 Individual Contribution
The individual contribution mark is calculated as follows:
First, each group member will assign a contribution score between 0 and 10 to every other
member in their group, reflecting their perception of each member’s effort and involvement
in the project.
Then, each student’s final individual contribution mark will be computed using the following
formula:
Individual Contribution Mark
=10× number of students in the group×
(
Total Marks Received by the individual
Total Marks Assigned by Group Members
)
For example, consider a group of four students. The following table represents their peer
evaluation results, where each row corresponds to the marks assigned by an individual, and
each column represents the marks received by each student.
Receiver
Total
A B C D
Grader
A nAB nAC nAD
B nBA nBC nBD
C nCA nCB nCD
D nDA nDB nDC
Total nA nB nC nD ntotal
Table 1: A table elaborating the calculation of individual contribution mark
Then the grade received by the ith student will be
10× 4× ni
ntotal
.
For example, marks received by Student A is 40 × nA
ntotal
. Under this marking scheme, a
student may receive an individual contribution mark greater than 10. Any excess marks will
be treated as bonus points and included in the final evaluation for this course.
This adjustment ensures that students who contribute more significantly to the project
receive appropriate recognition, while those who contribute less will have a lower individual
contribution mark.
7 Deadline
The deadline of the final project is Friday April 4th, 2025 23:59 PM.
6
