DEPARTMENT OF EDUCATIONAL TECHNOLOGY

Course Code : ICTE 4433

Course Title : Big Data and Analytics in Education
Assignment No : 03
Assignment On : When time series forecasting should be used? Explain it with
example.

Submitted To Submitted By
Sunjida Akter Md Ibna Shakil
Lecturer, Department of EdTech ID: 1902039

Bangladesh Sheikh Mujibur Rahman Educational Technology

Digital University, Bangladesh. Session: 2019-20

Date of Submission: 10th May, 2024

Question: When time series forecasting should be used? Explain it with example.
Answer:
Time series forecasting is a statistical technique used to predict future values based on historical
data points arranged in chronological order. It involves analyzing patterns and trends within the
time series data to make predictions about future values. Time series forecasting is commonly
applied in various fields such as finance, economics, weather forecasting, sales forecasting, and
many others where understanding and predicting future trends are crucial for decision-making.

The process of time series forecasting typically involves:

Data Collection: Gathering historical data points over a period of time, ensuring they are ordered
sequentially.

Exploratory Data Analysis (EDA): Analyzing the historical data to identify patterns, trends,
seasonality, and any other underlying structures.

Model Selection: Choosing an appropriate forecasting model based on the characteristics of the
time series data. Common models include ARIMA (AutoRegressive Integrated Moving Average),
Exponential Smoothing methods, and machine learning algorithms like LSTM (Long Short-Term
Memory) networks.

Model Training: Estimating the parameters of the chosen forecasting model using the historical
data.

Model Evaluation: Assessing the performance of the model using metrics such as Mean Absolute
Error (MAE), Mean Squared Error (MSE), or Root Mean Squared Error (RMSE) on a validation
dataset.

Forecasting: Using the trained model to generate predictions for future values of the time series.
Validation and Refinement: Validating the forecasts against actual data and refining the model if
necessary to improve accuracy.
Time series forecasting can be likened to peering through a rearview mirror in an attempt to
anticipate what lies ahead on the road. It's a technique applied in diverse domains to examine
historical data arranged chronologically, such as daily sales figures, monthly weather data, or
yearly stock prices, in order to make informed predictions about future patterns.

Now, when we're analyzing this data, we're essentially looking for a few key things:

Trend: Identifying trends involves recognizing the general direction in which the data is moving.
For instance, it could mean determining if sales are consistently rising over time or if there's a
gradual decrease in temperature each year. Understanding these trends gives us insight into the
broader patterns and tendencies within the data.

Seasonality: Consider the predictable cycles we observe annually, such as increased retail sales
during the holiday season or temperature variations with the changing seasons. These repetitive
patterns are known as seasonality.

Cyclical Patterns: Unlike seasonality, which follows a predictable yearly cycle, cyclical patterns
are longer-term trends that can be influenced by economic factors, like the boom-and-bust cycles of
the economy.

Irregular Events: Sometimes, unexpected things happen that don't follow any pattern. These
could be sudden spikes or drops in sales due to unforeseen events, like a viral social media post
driving up demand or a natural disaster disrupting supply chains.

Now, why do we bother with all this analysis? Well, because it helps us make better decisions in a
variety of areas:

Demand Forecasting: By predicting future demand for products or services, businesses can
manage their inventory more effectively and ensure they have enough stock to meet customer
needs without overstocking.

Financial Forecasting: Time series analysis is utilized by investors and financial institutions to
predict stock prices, currency exchange rates, and interest rates. This assists them in making well-
informed investment choices and effectively managing risks.

Economic Forecasting: Governments and policymakers use time series analysis to predict
economic indicators such as GDP growth, inflation rates, and unemployment rates, which inform
economic policies and business strategies.

Weather Forecasting: Meteorologists rely on time series analysis to predict weather patterns,
temperature fluctuations, and natural disasters, aiding in disaster preparedness, agriculture, and
energy management.

Healthcare Forecasting: Hospitals and healthcare providers leverage time series analysis to
predict patient admissions, disease outbreaks, and medication demand. This empowers them to
optimize resource allocation and service planning effectively.
Traffic Forecasting: Urban planners use time series analysis to predict traffic congestion, travel
times, and transportation demand, helping them design better transportation systems and
infrastructure.

Sales and Marketing Forecasting: Businesses use time series analysis to forecast sales revenue,
customer churn rates, and market trends, guiding their marketing strategies and revenue
projections.

Supply Chain Forecasting: Companies use time series analysis to predict supply chain demand,
production levels, and inventory management, ensuring smooth operations and efficient resource
allocation.

Population Forecasting: Governments and urban planners apply time series analysis to anticipate
population growth, demographic shifts, and trends in urbanization. This assists in making informed
decisions regarding housing, infrastructure development, and the provision of social services.

Essentially, time series analysis functions akin to a crystal ball, allowing us to glimpse into the
future by deciphering patterns from the past. It's a potent tool that empowers us to make wiser
decisions and enhance our readiness for forthcoming events.

Example: Let's take a sales dataset spanning a timeframe. On the horizontal axis, we depict
years, while on the vertical axis, we represent sales figures.

Fig: Sales time series