Skip to content
BENIBIX Learn

Time Series and Event Data

Level 3 · Kinds of data

Two kinds of data with a time on every record. Time series data comes on a steady beat. Event data is written each time something happens.

Course contentsLesson 8 of 11

After this lesson you can

  • Tell time series data from event data
  • Pick the right way to store and count each one
  • Spot when a gap in event data is not a zero

9 mincovers 2 glossary entries

Worth reading first: The Three Types of Data

How time series data and event data differ

Both kinds of data carry a time on every record. But they are made in two different ways. Time series data is read on a steady beat, such as a machine's memory use every 60 seconds. Event data is written each time something happens, such as a tap or a login. One has a clock behind it. The other has a deed behind it.

A time series is the same reading taken again and again. A smart meter reports power use every 15 minutes, which makes 96 readings a day. Each reading has the same fields: when, which meter and how much. The gaps between readings are fixed, so a missing one is easy to spot.

Time series data is usually summed or averaged over spans of time. What was the average each hour? What was the peak each day? Special databases keep it in time order and squeeze it hard, because the values change slowly. Old readings are often rolled up, so a year of minutes becomes a year of hours.

Event data records something that happened. Each event has a time, a name and its own details. A shop app might log viewed item, added to cart and paid. The gaps are uneven. A user might tap five times in two seconds and then nothing for a week.

Event data is counted and put in order. How many people paid today? What did a user do just before they gave up? Each event is a fact that already happened, so it is never changed. A refund is a new event, not an edit to the old payment.

People often read a gap in event data as a zero. No logins from a user this week might mean they left, or that the app stopped sending events. In a time series, a missing reading is a clear fault. In event data, silence can mean nothing happened, or that the pipe broke.

Which one you have

  1. 1

    Time series data

    You have it when a sensor or a system reports the same measure on a clock. Keep it in time order, and roll old readings up.

  2. 2

    Event data

    You have it when each record is something a person or a machine did. Keep every event as it was, and count them later.

A time series ticks on a clock. Events land whenever something happens, so their gaps tell you nothing on their own.

Worked example

A gym and its two kinds of data

Jack's gym logs an event each time a member swipes in, about 900 a day. It also reads the air in the main room every 5 minutes.

The swipes show 40 members stopped coming in March. The air readings show one sensor sent nothing for 3 days, which was a fault, not a quiet room.

The words for this lesson2

Check yourself

A smart meter sends a reading every 15 minutes. What kind of data is this?

Finished this lesson?

Marking it complete also marks its 2 glossary entries learned, which updates your roadmap and your saved page. 0 are already marked.