Page 214 - AI Computer 10

Error quantifies how far off the prediction was from reality, whereas accuracy quantifies the frequency with
which the model makes correct prediction.

Let us calculate the accuracy of the House Price prediction AI model using the values shown in the table given
below.

Predicted House Actual House Error = Absolute Error Rate Accuracy Rate = 1- Accuracy (%) =
Price (in `) Price (in `) (Actual – Predicted) = Error/ Error Rate Accuracy Rate *
Actual 100

7200000 – 7000000 1 – 0.0285 =
7200000 7000000 0.02857 97.15
= 200000 0.9715
5600000 – 5500000 1 – 0.0181 =
5600000 5500000 0.01818 98.19
= 100000 0.9819
4550000 – 4400000 1 – 0.0329 =
4400000 4550000 0.03296 96.71
= 150000 0.9671
8650000 – 8400000 1 – 0.0289 =
8400000 8650000 0.0289 97.11
= 250000 0.9711
Note, that Absolute means the difference between the two values, irrespective of the negative sign.

CLASSIFICATION MODEL EVALUATION TERMINOLOGIES OR METRICS
In general, Model Evaluation stands on the two pillars of accuracy and error. Let’s explore various terminologies/
metrics used in Model evaluation with the help of a scenario.

u Scenario: Suppose you have been given a task to deploy an AI based prediction model in an area which is
prone to floods. The objective of the model is to assess the area and predict whether a flood has occurred
or not. In this case, two terms or conditions are used to determine the efficiency of the model. These terms
are:
u Prediction: Prediction is the output generated by the model.

u Reality: Reality is the real condition of the area when the prediction has been made.
There arise four possibilities regarding the model. These are:

Case I: Did a flood occur here?

Prediction : Yes True Positive Reality : Yes

Here, the positive prediction of a flood made by the model matches the reality of a flood. This is called a True
Positive (TP).

80
80