Time Series: Trends, Cycles, Seasonality & Forecasting

TAKEAWAYS:

• Time series data shows how a variable (like sales) changes over time, revealing trends (long-term growth/decline), cycles (multi-year fluctuations), and seasonality (repeating short-term patterns).
• Moving averages smooth out random fluctuations to highlight trends—simple moving average (SMA) and weighted moving average (WMA) are key tools.
• Decomposition breaks a time series into four components: Trend (T), Seasonal (S), Cyclical (C), and Irregular (I) using additive or multiplicative models.
• Forecasting uses past data to predict future values; methods include naïve forecasting, exponential smoothing, and linear regression for trend analysis.
• Businesses use time series to plan inventory, budgets, and marketing strategies by anticipating demand swings.
• NEB exams test calculations (e.g., moving averages, seasonal indices) and interpretation (e.g., identifying trends from graphs).

---

---

What is a Time Series?

A time series is a collection of data points recorded at regular intervals over time. Examples:

• Monthly sales of a company for 5 years.
• Quarterly GDP growth of Nepal.
• Daily temperature readings in Kathmandu.

Why Study Time Series?

Businesses and economists use time series to:

1. Understand past patterns (e.g., why sales spike in December).
2. Predict future values (e.g., next year’s sales).
3. Make data-driven decisions (e.g., stock inventory, marketing campaigns).

---

Caption: Monthly sales data showing an upward trend with seasonal spikes at year-end.

---

Components of a Time Series

Every time series has four components:

1. Trend (T): Long-term increase or decrease (e.g., sales growing over 5 years).
2. Seasonal (S): Repeating short-term patterns (e.g., higher sales in December).
3. Cyclical (C): Medium-term fluctuations (e.g., economic booms and recessions).
4. Irregular/Random (I): Unexpected shocks (e.g., a natural disaster reducing sales).

Additive vs. Multiplicative Models

---

1. Moving Averages (Smoothing Data)

Moving averages help remove short-term fluctuations to reveal the underlying trend.

Types of Moving Averages

Worked Example: Simple Moving Average (SMA)

Data: Quarterly sales (in thousands) for 2021–2022:

Task: Calculate the 3-quarter SMA for Q2 2021 to Q2 2022.

Solution:

1. For Q2 2021 (t=2): ${\text{SMA}}_2=\frac{50+55+60}{3}=\frac{165}{3}=55$
2. For Q3 2021 (t=3): ${\text{SMA}}_3=\frac{55+60+65}{3}=\frac{180}{3}=60$
3. For Q4 2021 (t=4): ${\text{SMA}}_4=\frac{60+65+70}{3}=\frac{195}{3}=65$
4. For Q1 2022 (t=5): ${\text{SMA}}_5=\frac{65+70+75}{3}=\frac{210}{3}=70$

Graphical Representation: Caption: SMA smooths out fluctuations, revealing the upward trend.

---

2. Decomposition of Time Series

Decomposition separates the trend (T), seasonal (S), cyclical (C), and irregular (I) components.

Steps for Additive Decomposition

1. Detrend the data (remove trend using moving averages).
2. Remove seasonality (average seasonal effects).
3. Calculate seasonal indices.
4. Reconstruct the series.

Worked Example: Seasonal Decomposition

Data: Quarterly sales (in thousands) for 3 years:

Step 1: Calculate Trend (3-year SMA)

• Use centered moving average (CMA) for Q2 2022 and Q2 2023.
• For Q2 2022: $\text{CMA}=\frac{50+55+60+65+70+75}{6}=62.5$

Step 2: Detrend the Data Subtract trend from actual sales:

Step 3: Calculate Seasonal Indices Average detrended values for each quarter:

• Q1 Seasonal Index = $\frac{-5+(-2.5)+0}{3}=-2.5$
• Similarly, calculate for Q2, Q3, Q4.

Step 4: Adjust for Seasonality Add seasonal index back to trend to forecast:

• Forecast for Q1 2024 = Trend (2023 Q4) + Q1 Seasonal Index = 90 + (-2.5) = 87.5

---

3. Forecasting Methods

A. Naïve Forecasting

Assumes the next value = last observed value.

• Pros: Simple, no calculations.
• Cons: Ignores trends/seasonality.

Example: If last quarter’s sales = 90, then next quarter forecast = 90.

B. Exponential Smoothing

Gives more weight to recent data using a smoothing factor ($\alpha$). Formula: $F_{t+1}=\alpha Y_t+(1-\alpha )F_t$

• $\alpha$ = smoothing factor (0 < $\alpha$ < 1)
• $F_t$ = forecast at time $t$

Worked Example: Given $\alpha =0.3$, and sales data:

Forecast for Q5: $F_5=0.3\times 65+0.7\times 54.4=57.08$

C. Linear Regression for Trend Forecasting

If the trend is linear, use: $Y=a+bX$

• $Y$ = dependent variable (sales)
• $X$ = independent variable (time)
• $a$ = intercept, $b$ = slope

Worked Example:

Step 1: Calculate Slope ($b$) $b=\frac{n\sum XY-\sum X\sum Y}{n\sum X^2-(\sum X{)}^2}$ $=\frac{3\times 510-6\times 180}{3\times 14-36}=\frac{1530-1080}{42-36}=\frac{450}{6}=10$

Step 2: Calculate Intercept ($a$) $a=\frac{\sum Y-b\sum X}{n}=\frac{180-10\times 6}{3}=40$

Forecast Equation: $Y=40+10X$ Forecast for Year 4: $Y=40+10\times 4=80$

---

4. Applications of Time Series in Business

---

5. Advantages and Limitations

---

Exam Tip: How to Score Full Marks in NEB

1. Understand the Question Type

   • Calculation-based: Moving averages, seasonal indices, forecasts.
   • Theoretical: Define components (trend, seasonality, cycles).
   • Interpretation: Explain trends from graphs.

2. Show All Steps

   • NEB marks working, not just the final answer.
   • Example: For SMA, write each term clearly.

3. Use Graphs

   • Always plot data if asked to "analyze trends."
   • Label axes, trends, and seasonal patterns.

4. Common Mistakes to Avoid

   • Ignoring units (e.g., sales in NRs, not just numbers).
   • Misapplying formulas (e.g., using SMA for odd periods without centering).
   • Forgetting to interpret (e.g., "The trend is increasing" after forecasting).

5. Practice NEB-Style Questions Example Question (NEB Board Style):

   "The quarterly sales data of a company is given below. Calculate the 4-quarter centered moving average and comment on the trend."

   Year	Q1	Q2	Q3	Q4
   2021	100	120	110	150
   2022	130	140	160	180

   Solution Steps:

   1. Calculate 4-quarter CMA (average of Q1–Q4, Q2–Q5, etc.).
   2. Plot the data and CMA.
   3. Comment: "The trend is upward, indicating growing sales."

---

Summary Checklist

Before the exam, ensure you can: ✅ Calculate SMA, WMA, and CMA. ✅ Decompose a time series into trend, seasonal, cyclical, and irregular components. ✅ Forecast using naïve method, exponential smoothing, and linear regression. ✅ Interpret graphs (identify trends, seasonality). ✅ Apply time series to real-world business problems.

---

Final Note: Time series is 80% calculations + 20% interpretation. Practice past NEB papers to master both! 🚀