Sample Machine Learning problem

Let’s understand ML better by analyzing the weather problem. Assume that you want to play soccer on the weekend. Consider this small data set of various weather conditions.

Problem statement

Based on the dataset of weather conditions, predict whether you will be able to play soccer on the weekend.

Data set

Outlook	Temperature	Humidity	Windy	Play
Sunny	Hot	High	False	No
Rainy	Mild	High	False	Yes
Sunny	Cool	Normal	False	Yes

Features/attributes

Based on the data set, the features (also known as attributes) of this problem are:

Outlook
Temperature
Humidity
Windy

This information is called ‘feature’ or ‘attribute’ because it helps to determine which weather conditions are favorable for playing soccer on a specific day.

Target variable/Outcome

Based on the data set, the target variable (also known as outcome) is ‘Play’. You are required to predict what the outcome (also known as target) will be i.e. you must be able to determine whether you will be able to play the game.

Vector

A vector is a row in the data set. For example, in the data set for this problem Sunny, Hot, High, False, No is a vector. No is the target for this vector.

ML model

An ML model is a model where an algorithm learns patterns from the given data set (known as training data set) to be able to make predictions on a new data set (known as test data set).

Evaluation metrics

After a model is trained, the next step is to evaluate the model’s performance i.e. to check how accurately the model has learned from the data. This performance is determined by using mathematical formulas.

Typical learning process

The process that is used by an ML algorithm to learn from the data is as follows:

S. No.	Step	Description
1	Acquire data	The data set is loaded in the software (R, Python etc.).
2	Clean data	The data is cleaned (i.e. spaces are removed, missing data is input etc.) to make it ready for the next step.
3	Explore data	The behavior of the given features is studied by using charts and graphs.
4	Feature engineering	Using the information from the previous step, new features are created (if possible).
5	Model training	The model is trained by using a suitable ML algorithm.
6	Model evaluation	The performance of the resulting trained model is evaluated.
7	Model testing	The model is set to predict on a new data set and it is evaluated again.

Articles in this section