Probabilities in Decision Models

Decoding Decision Modeling

Tiago Brasil

Lead Data Engineer

Probabilistic Models

A white rectangle with a white outline to facilitate the slide formating Probabilistic Models support decision-making by providing a structured approach to handle uncertainties.

By applying probabilities in decision models we can:

  • Quantify uncertainties by assigning values between 0 and 1 to express the likelihood of outcomes

A diagram centered around a target symbol, representing decision-making. Four arrows point outward from the target in four different directions. Each arrow leads to a label and an icon of Objectives, Alternatives, Constraints and Uncertainties (Highlighted in yellow with the icon of a ruler pointing to it)

Decoding Decision Modeling

Probabilities and uncertainties

Probabilities allow decision-makers to quantify uncertainties by assigning numerical values to the likelihood of different outcomes occurring.

Mathematical definition of probability equals to favorable outcomes divided by number of possible outcomes

Probability diagram illustrating different levels of uncertainties given the probability.

Decoding Decision Modeling

Probabilities in a coin toss

In a standard coin with heads on one side, and tails on the other, let's find the likelihood of the coin landing on heads twice consecutively.

Superior part of a decision tree of a coin toss as a root node and head and tail as the two potential outcomes

Decoding Decision Modeling

Probabilities in a coin toss

In a standard coin with heads on one side, and tails on the other, let's find the likelihood of the coin landing on heads twice consecutively.

A decision tree with three layers. Layer 1 is the root node representing the coin toss, layer two is heads and tail as the outcomes of layer two and layer 3 is head and tail for each possible outcome of layer 2.

Decoding Decision Modeling

Probabilities in a coin toss

In a standard coin with heads on one side, and tails on the other, let's find the likelihood of the coin landing on heads twice consecutively.

A decision tree representing a consecutive coin toss with all the possible outcomes and probabilities of each one highlighted in yellow

Decoding Decision Modeling

Probabilities in a coin toss

In a standard coin with heads on one side, and tails on the other, let's find the likelihood of the coin landing on heads twice consecutively.

A decision tree highlighting the outcome of the first coin toss as head, as well as the second outcome and the probability of this happening.

Decoding Decision Modeling

Probabilities in a coin toss

A white rectangle with a white outline to facilitate the slide formating A decision tree highlighting the outcome of the first coin toss as head, the second outcome as tail and the probability of this happening.

A label showing that the probability of getting heads first and tails second is 25% 0.5 times 0.5

Decoding Decision Modeling

Probabilities in a coin toss

A white rectangle with a white outline to facilitate the slide formating A decision tree highlighting tail as the outcomes of first and second coin toss and the probability of this happening.

A label showing that the probability of getting one tails from two consecutive coin flips is 75% 0.25 plus 0.25 plus 0.25

Decoding Decision Modeling

Building a Probabilistic Decision Model

The marketing team of a retail company is deciding where to launch the next campaign.

The beginning of a decision tree with marketing campaign as the root node and social media and newspaper ads as alternatives

Decoding Decision Modeling

Building a Probabilistic Decision Model

The marketing team of a retail company is deciding where to launch the next campaign.

The beginning of a decision tree with marketing campaign as the root node, social media and newspaper ads as alternatives and a $100 label representing the cost for each

Decoding Decision Modeling

Building a Probabilistic Decision Model

The marketing team of a retail company is deciding where to launch the next campaign.

A decision tree with marketing campaign as the root node and social media and newspaper ads as alternatives. Each alternative has success represented as a green triangle and fail represented as a red triangle

Decoding Decision Modeling

Building a Probabilistic Decision Model

The marketing team of a retail company is deciding where to launch the next campaign.

A decision tree with marketing campaign as the root node and social media and newspaper ads as alternatives. Each alternative has success represented as a green triangle and fail represented as a red triangle. The probability of each of these outcomes is labeled as 50%

Decoding Decision Modeling

Building a Probabilistic Decision Model

The marketing team of a retail company is deciding where to launch the next campaign.

A decision tree with marketing campaign as the root node and social media and newspaper ads as alternatives. Each alternative has success represented as a green triangle and fail represented as a red triangle. The profit or loss value for each outcome is highlighted in yellow

Decoding Decision Modeling

Building a Probabilistic Decision Model

The marketing team of a retail company is deciding where to launch the next campaign.

A decision tree with marketing campaign as the root node and social media and newspaper ads as alternatives. Each alternative has success represented as a green triangle and fail represented as a red triangle. The profit or loss value for each outcome is highlighted in yellow. An arrow with a subtraction sign pointing to the cost.

Decoding Decision Modeling

Building a Probabilistic Decision Model

The marketing team of a retail company is deciding where to launch the next campaign.

The final decision tree of the problem with the alternatives, the outcomes for each alternative, their probabilities and the updated profit or loss value for each outcome

Decoding Decision Modeling

Evaluating expected value

A white rectangle with a white outline to facilitate the slide formating

The final decision tree of the problem with the alternatives, the outcomes for each alternative, their probabilities and the updated profit or loss value for each outcome

A white rectangle with a white outline to facilitate the slide formating A label indicating that expected value is predicted success times potential earnings plus predicted failure times potential loss

Decoding Decision Modeling

Evaluating expected value

A white rectangle with a white outline to facilitate the slide formating A label indicating the calculus of the expected value for the social media alternative being $300

A white rectangle with a white outline to facilitate the slide formating A label indicating that expected value is predicted success times potential earnings plus predicted failure times potential loss

Decoding Decision Modeling

Evaluating expected value

A white rectangle with a white outline to facilitate the slide formating A label indicating the calculus of the expected value for the newspaper ads alternative being $335

A white rectangle with a white outline to facilitate the slide formating A label indicating that expected value is predicted success times potential earnings plus predicted failure times potential loss

Decoding Decision Modeling

Evaluating expected value

A white rectangle with a white outline to facilitate the slide formating A label indicating the calculus of the expected value for the social media alternative being $300

A white rectangle with a white outline to facilitate the slide formating A label indicating the calculus of the expected value for the newspaper ads alternative being $335

Decoding Decision Modeling

Let's practice!

Decoding Decision Modeling

Preparing Video For Download...