Time Series • Statistics Background for Forecasting
Time Series / Numerical Description of Time Series Data

Numerical Description of Time Series Data

Notes 3 Statistics Background for Forecasting

Numerical description uses summary numbers to describe a set of observations: typical size, spread, and extreme values. These measures help compare series and check whether a value is unusually large or small.

Notes

Numerical Description of Time Series Data

Definition

Numerical description uses summary numbers to describe a set of observations: typical size, spread, and extreme values. These measures help compare series and check whether a value is unusually large or small.

Example

Five weekly sales figures 10, 12, 11, 13, 14 can be summarised by a mean of 12 packets and a range of 4 packets.

Why Numerical Measures Are Used

A plot shows shape. Numbers answer questions such as: What is a typical value? How much do values vary? What were the lowest and highest observations? Both views are needed. Numerical summaries ignore time order unless they are calculated on specially chosen periods, so they cannot replace a time series plot.

Small Dataset

Weekly packet sales: 10, 12, 11, 13, 14 (already in time order).

Main Measures

Measure Meaning Interpretation
Mean Arithmetic average of all observations Typical level of the series
Median Middle value after sorting Typical level that is less affected by one extreme
Minimum / Maximum Smallest and largest observations The observed extremes
Range Maximum minus minimum Overall spread of the recorded values
Variance Average of squared deviations from the mean How widely values scatter; units are squared
Standard deviation Square root of variance Spread in the same units as the data
Quartiles Values that split the ordered data into four parts Where the lower, middle and upper observations lie

Calculations for the Example

Mean = (10 + 12 + 11 + 13 + 14) / 5 = 60 / 5 = 12.

Ordered values: 10, 11, 12, 13, 14. Median = 12.

Minimum = 10, maximum = 14, range = 14 − 10 = 4.

Deviations from the mean 12: −2, 0, −1, 1, 2. Squared deviations: 4, 0, 1, 1, 4. Sum = 10. Sample variance with divisor n − 1 = 10 / 4 = 2.5. Sample standard deviation = √2.5, which is a little more than 1.5.

The lower quartile of these five points is near 11 and the upper quartile is near 13. With only five observations, quartiles are a rough split, not a precise population claim.

Practical Interpretation

The mean and median both equal 12, so the typical weekly sales in this tiny sample sit at 12 packets. The range of 4 packets shows modest spread. None of these numbers, by themselves, tell you that week 1 came before week 5. That order is seen on the time plot.

Exam Note

Numerical summaries cannot reveal all time-order patterns. A time series plot is also important.

Small Educational Python Example

# Import libraries import pandas as pd sales = pd.Series([10, 12, 11, 13, 14], name="sales") print("Mean:", sales.mean()) print("Median:", sales.median()) print("Min, max:", sales.min(), sales.max()) print("Range:", sales.max() - sales.min()) print("Sample variance:", sales.var()) print("Sample standard deviation:", sales.std()) print(sales.describe())

The printed mean and median should match 12, and the range should be 4. Software may show extra decimal places for variance and standard deviation; use the same formula convention (sample or population) that your course requires.

Exam-Oriented Key Points

  1. Mean and median describe typical level; variance and standard deviation describe spread.
  2. Minimum, maximum and range describe extremes.
  3. Quartiles split ordered data into four parts.
  4. Always read numerical summaries together with a time plot.