Skip to content
Glacius
ProbabilityConcept reference

Event probabilities

For a finite sample space, an event probability is the sum of its distinct outcome masses.

On this page 7 sections
  1. Overview
  2. An event’s probability is the total mass of its outcomes
  3. If a model returns exactly one label, a group of labels is an event
  4. Counting favorable outcomes works only when every outcome has equal mass
  5. Key takeaway
  6. Sources & further reading
  7. Concept connections

01An event’s probability is the total mass of its outcomes#

An event’s probability is the total mass of its outcomes. Distinct outcomes are alternatives for one trial, so their masses add.

Let A, B, C have masses 0.15, 0.5, 0.35. For E = {A, C}, select 0.15 and 0.35.

The top probability whole consists of A=.15, B=.5, C=.35. The lower strip includes only A and C on the same absolute scale, totaling .5.The top probability whole consists of A=.15, B=.5, C=.35. The lower strip includes only A and C on the same absolute scale, totaling .5.
Figure 1The top probability whole consists of A=.15, B=.5, C=.35. The lower strip includes only A and C on the same absolute scale, totaling .5.
Link to this figure ↗Download SVGDownload PNG
Check your reasoning

Masses: A=0.2, B=0.5, C=0.3. Event E = {A, C}. Find P(E).

Show answer and explanation
0.5

Add the masses of A, C: 0.5.

02If a model returns exactly one label, a group of labels is an event#

If a model returns exactly one label, a group of labels is an event. Add their supplied masses. The labels need not have equal chances.

Check your reasoning

A model returns one label. Masses: red=0.4, amber=0.35, green=0.25. An alert uses red or amber. Find its probability.

Show answer and explanation
0.75

Add the alert labels’ masses: 0.75.

03Counting favorable outcomes works only when every outcome has equal mass#

Counting favorable outcomes works only when every outcome has equal mass. With unequal masses, use the weights. Writing a label twice does not create a second outcome.

Check your reasoning

Masses: L=0.1, M=0.2, N=0.7. Event E = {L, M}. A learner gives P(E)=2/3. Repair it.

  1. A2/3
  2. B1/3
  3. C0.3
Show answer and explanation
0.3

0.1 + 0.2 = 0.3.

Key takeaway

Select the event’s outcomes and add their masses once each. Use counts only when the underlying outcomes are equally likely.

Sources & further reading

  1. [1]

Reference this concept

Link to this page, a section, or an individual figure.

Glacius. “Event probabilities.” Math behind ML. /learn/p-event-sum