Page 145 - Ai_C10_Flipbook
P. 145
Here,
• The whole dataset is divided into training and testing data.
• The ML algorithm is trained using the training data. This involves feeding the data into the algorithm, which
learns patterns and relationships to create a model.
• The trained model is tested or applied using testing data, ensuring that the model generalises well to unseen
data.
• The ML algorithm processes the input data and provides a prediction or result. This is the model’s output
• The predictions are evaluated on the basis of following criteria:
o If they are accurate or meet the desired expectations (indicated by a thumbs-up), the model is
considered successful.
o If the predictions are unsatisfactory (indicated by a thumbs-down), adjustments are made. This
could involve refining the training data, tuning the model’s parameters, or retraining the algorithm.
Evaluation Techniques
Evaluation techniques involve assessing a machine learning model’s performance on training and test data. The
three types of evaluation techniques are given below:
Overfitting Underfitting Perfect Fit
(forcefitting- too good to be true) (too simple to explain) (captures data trends and
generalises well.)
The description of these evaluation models is as follows:
1. Overfitting Model: The model (red curve) fits the training data perfectly, including noise, but performs poorly
on the testing data, leading to poor generalisation. In overfitting, the model is too complex and performs well
on training data but poorly on test data. It has low bias and high variance. The model memorizes the training
data but struggles to generalise to new, unseen data.
2. Underfitting Model: The model (purple line) is too simplistic, failing to capture the pattern in both the training
and testing data. It has high bias and low variance. The model fails to capture the underlying patterns in the
data.
3. Perfect Fit Model: The model (green curve) balances complexity and generalisation, fitting the training data
well and performing well on the testing data. It performs well on both training and test data and generalises
effectively to new data.
Evaluating Models 143

